Datasets:

sswt

/

arxiv_sample_50K

like 0

Large mass splittings for fourth generation fermions allowed by LHC Higgs exclusion ; In the context of the Standard model with a fourth generation, we explore the allowed mass spectra in the fourth generation quark and lepton sectors as functions of the Higgs mass. Using the constraints from unitarity and oblique parameters, we show that a heavy Higgs allows large mass splittings in these sectors, opening up new decay channels involving W emission. A Higgs heavier than sim 800 GeV would in fact necessitate either a heavy quark decay channel t' b'Wb' t' W or a heavy lepton decay channel tau' nu' W as long as the mixing between the third and fourth generations is small. This mixing tends to suppress the mass splittings and hence the Wemission channels. The possibility of the Wemission channel could substantially change the search strategies of fourth generation fermions at the LHC and impact the currently reported mass limits.

Collapsing shearfree perfect fluid spheres with heat flow ; A global view is given upon the study of collapsing shearfree perfect fluid spheres with heat flow. We apply a compact formalism, which simplifies the isotropy condition and the condition for conformal flatness. This formalism also presents the simplest possible version of the main junction condition, demonstrated explicitly for conformally flat and geodesic solutions. It gives the right functions to disentangle this condition into well known differential equations like those of Abel, Riccati, Bernoulli and the linear one. It yields an alternative derivation of the general solution with functionally dependent metric components. We bring together the results for static and time dependent models to describe six generating functions of the general solution to the isotropy equation. Their common features and relations between them are elucidated. A general formula for separable solutions is given, incorporating collapse to a black hole or to a naked singularity.

Generalized de Sitter Space in ndimensional Minkowski Space ; In this paper, we generalize the defining equation for de Sitter space by replacing the de Sitter radius with a function f satisfying certain conditions; each resulting hypersurface is diffeomorphic to de Sitter space, and has a geometry and causal character which is controlled by the choice of f. Necessary and sufficient conditions are obtained for a hypersurface to be timelike, null, or spacelike in the generalized model; in the nonnull case, the geometry is given by a warped product. Several examples of timelike, null, and spacelike hypersurfaces are presented. Lastly, we calculate the Ricci tensor and scalar curvature for a special family of 4dimensional generalized de Sitter spaces.

Automated Word Puzzle Generation via Topic Dictionaries ; We propose a general method for automated word puzzle generation. Contrary to previous approaches in this novel field, the presented method does not rely on highly structured datasets obtained with serious human annotation effort it only needs an unstructured and unannotated corpus i.e., document collection as input. The method builds upon two additional pillars i a topic model, which induces a topic dictionary from the input corpus examples include e.g., latent semantic analysis, groupstructured dictionaries or latent Dirichlet allocation, and ii a semantic similarity measure of word pairs. Our method can i generate automatically a large number of proper word puzzles of different types, including the odd one out, choose the related word and separate the topics puzzle. ii It can easily create domainspecific puzzles by replacing the corpus component. iii It is also capable of automatically generating puzzles with parameterizable levels of difficulty suitable for, e.g., beginners or intermediate learners.

Plane waves in the generalized Galileon theory ; We present an exact plane wave solution of the most general shiftsymmetric Horndeski generalized Galileon theory. The solution consists of the scalar part, and the gravitational part with two polarization modes. The former is due to the presence of the nontrivial Galileon scalar field, and it is parametrized by an arbitrary function of the lightcone coordinate. For a trivial scalar field configuration the solution is equivalent to the plane gravitational wave in General Relativity. When the metric is Minkowski, we reproduce known results for the plane waves of kessence and a solitonlike solutions of a noncovariant Galileon model in a flat spacetime.

A Generalized MeanReverting Equation and Applications ; Consider a meanreverting equation, generalized in the sense it is driven by a 1dimensional centered Gaussian process with Holder continuous paths on 0,T T 0. Taking that equation in rough paths sense only gives local existence of the solution because the nonexplosion condition is not satisfied in general. Under natural assumptions, by using specific methods, we show the global existence and uniqueness of the solution, its integrability, the continuity and differentiability of the associated Ito map, and we provide an Lpconverging approximation with a rate of convergence pgeq 1. The regularity of the Ito map ensures a large deviation principle, and the existence of a density with respect to the Lebesgue measure, for the solution of that generalized meanreverting equation. Finally, we study a generalized meanreverting pharmacokinetic model.

A simple algorithm for automatic Feynman diagram generation ; An algorithm for the automatic Feynman diagram FD generation is presented in this paper. The algorithm starts directly from the definition formula of FD, and is simple in concept and easy for coding. The symmetry factor for each FD is naturally generated. It is expected to bring convenience for the researchers who are studying new calculation techniques or making new calculation tools and for the researchers who are studying effective field theory. A Cprogram made from the algorithm is also presented, which is short, fast, yet very general purpose it receives arbitrary user defined model and arbitrary process as input and generates FD's at any order.

External Memory based Distributed Generation of Massive Scale Social Networks on Small Clusters ; Small distributed systems are limited by their main memory to generate massively large graphs. Trivial extension to current graph generators to utilize external memory leads to large amount of random IO hence do not scale with size. In this work we offer a technique to generate massive scale graphs on small cluster of compute nodes with limited main memory. We develop several distributed and external memory algorithms, primarily, shuffle, relabel, redistribute, and, compressedsparserow csr convert. The algorithms are implemented in MPIpthread model to help parallelize the operations across multicores within each core. Using our scheme it is feasible to generate a graph of size 238 nodes scale 38 using only 64 compute nodes. This can be compared with the current scheme would require at least 8192 compute node, assuming 64GB of main memory. Our work has broader implications for external memory graph libraries such as STXXL and graph processing on SSDbased supercomputers such as Dash and Gordon 12.

A Survey on Techniques of Improving Generalization Ability of Genetic Programming Solutions ; In the field of empirical modeling using Genetic Programming GP, it is important to evolve solution with good generalization ability. Generalization ability of GP solutions get affected by two important issues bloat and overfitting. We surveyed and classified existing literature related to different techniques used by GP research community to deal with these issues. We also point out limitation of these techniques, if any. Moreover, the classification of different bloat control approaches and measures for bloat and overfitting are also discussed. We believe that this work will be useful to GP practitioners in following ways i to better understand concepts of generalization in GP ii comparing existing bloat and overfitting control techniques and iii selecting appropriate approach to improve generalization ability of GP evolved solutions.

Nonlinear superhorizon curvature perturbation in generic singlefield inflation ; We develop a theory of nonlinear cosmological perturbations on superhorizon scales for generic singlefield inflation. Our inflaton is described by the Lagrangian of the form WX,phiGX,phiBoxphi with Xpartialmuphipartialmuphi2, which is no longer equivalent to a perfect fluid. This model is more general than kinflation, and is called Ginflation. A general nonlinear solution for the metric and the scalar field is obtained at second order in gradient expansion. We derive a simple master equation governing the largescale evolution of the nonlinear curvature perturbation. It turns out that the nonlinear evolution equation is deduced as a straightforward extension of the corresponding linear equation for the curvature perturbation on uniform phi hypersurfaces.

Radiative generation of the Higgs potential ; We consider the minimal extension of the Standard Model with a generalized BL gauge symmetry U1X for generating the Higgs potential radiatively. Assuming that the full scalar potential vanishes at the vacuum instability scale, we achieve the goal in terms of two free parameters, the X gauge coupling and the righthanded neutrino Yukawa coupling. The X gauge symmetry is broken spontaneously by the ColemanWeinberg mechanism while the scale symmetry breakdown induces electroweak symmetry breaking through the radiative generation of appropriate scalar quartic couplings. We show that there is a reasonable parameter space that is consistent with a correct electroweak symmetry breaking and the observed Higgs mass.

Quantum cosmology in HoravaLifshitz gravity ; Quantum cosmology is studied within the framework of the minimal quantum gravity theory proposed by Hovrava. For this purpose we choose the KantowskiSachs KS model and construct the corresponding WheelerDeWitt equation. We study the solution to this equation in the ultraviolet limit for different values of the running parameter lambda of the theory. It is observed that the wave packet for this Universe changes completely compared with the one observed in the infrared general relativity regime. We also look at the classical solutions by means of a WKB semiclassical approximation. It is observed that if lambda takes its relativistic value lambda 1 a generalized KS metric is obtained which differs from the usual KS solution in general relativity by an additional term arising from the higherorder curvature terms in the action and which dominates the behavior of the solution for very small values of the time parameter. We discuss the physical properties of this solution by comparing it with the usual KS solution in general relativity. The resulting solution has no horizons but singularities.

Generalized Langevin equation description of the stochastic oscillations of general relativistic disks ; We consider a description of the stochastic oscillations of the general relativistic accretion disks around compact astrophysical objects based on the generalized Langevin equation, which accounts for the general retarded effects of the frictional force, and on the fluctuationdissipation theorems. The vertical displacements, velocities and luminosities of the stochastically perturbed disks are explicitly obtained for both the Schwarzschild and the Kerr cases. The Power Spectral Distribution of the luminosity it is also obtained, and it is shown that it has nonstandard values. The theoretical predictions of the model are compared with the observational data for the luminosity time variation of the BL Lac S5 0716714 object.

Almost Kaehler Ricci Flows and Einstein and LagrangeFinsler Structures on Lie Algebroids ; In this work we investigate Ricci flows of almost Kaehler structures on Lie algebroids when the fundamental geometric objects are completely determined by semi Riemannian metrics, or effective regular generating Lagrange Finsler, functions. There are constructed canonical almost symplectic connections for which the geometric flows can be represented as gradient ones and characterized by nonholonomic deformations of Grigory Perelman's functionals. The first goal of this paper is to define such thermodynamical type values and derive almost Kahler Ricci geometric evolution equations. The second goal is to study how fixed Lie algebroid, i.e. Ricci soliton, configurations can be constructed for Riemannian manifolds andor co tangent bundles endowed with nonholonomic distributions modelling generalized Einstein or Finsler Cartan spaces. Finally, there are provided some examples of generic offdiagonal solutions for Lie algebroid type Ricci solitons and effective Einstein and LagrangeFinsler algebroids.

Generalized massive gravity in arbitrary dimensions and its Hamiltonian formulation ; We extend the fourdimensional de RhamGabadadzeTolley dRGT massive gravity model to a general scalar massivetensor theory in arbitrary dimensions, coupling a dRGT massive graviton to multiple scalars and allowing for generic kinetic and mass matrix mixing between the massive graviton and the scalars, and derive its Hamiltonian formulation and associated constraint system. When passing to the Hamiltonian formulation, two different sectors arise a general sector and a special sector. Although obtained via different ways, there are two second class constraints in either of the two sectors, eliminating the BD ghost. However, for the special sector, there are still ghost instabilities except for the case of two dimensions. In particular, for the special sector with one scalar, there is a second BD ghost.

Reflected Generalized Beta Inverse Weibull Distribution definition and properties ; In this paper we study a broad class of distribution functions which is defined by means of reflected generalized beta distribution. This class includes that of Betagenerated distribution as a special case. In particular, we use this class to extend the Inverse Weibull distribution in order to obtain the Reflected Generalized Beta Inverse Weibull Distribution.For this new distribution, moments, entropy, order statistics and a reliability measure are derived. The link between the Inverse Weibull and the Dagum distribution is generalized. Then the maximum likelihood estimators of the parameters are examined and the observed Fisher information matrix provided. Finally, the usefulness of the model is illustrated by means of an application to real data.

Degrees of Freedom of Generic BlockFading MIMO Channels without A Priori Channel State Information ; We studynthe highSNR capacity of generic MIMO Rayleigh blockfading channels in the noncoherent setting where neither transmitter nor receiver has a priori channel state information but both are aware of the channel statistics. In contrast to the wellestablished constant blockfading model, we allow the fading to vary within each block with a temporal correlation that is generic in the sense used in the interferencealignment literature. We show that the number of degrees of freedom of a generic MIMO Rayleigh blockfading channel with T transmit antennas and block length N is given by T11N provided that TN and the number of receive antennas is at least TN1NT. A comparison with the constant blockfading channel where the fading is constant within each block shows that, for large block lengths, generic correlation increases the number of degrees of freedom by a factor of up to four.

Local Tb theorem with L2 testing conditions and general measures CalderonZygmund operators ; Local Tb theorems with Lp type testing conditions, which are not scale invariant, have been studied widely in the case of the Lebesgue measure. Until very recently, local Tb theorems in the nonhomogeneous case had only been proved assuming scale invariant Linfty or BMO testing conditions. The combination of nonscaleinvariance and general measures is a delicate issue. In a previous paper we overcame this obstacle in the model case of square functions defined using general measures. In this paper we finally tackle the very demanding case of Calder'onZygmund operators. That is, we prove a nonhomogeneous local Tb theorem with L2 type testing conditions for all Calder'onZygmund operators. In doing so we prove general twisted martingale transform inequalities which turn out to be subtle in our general framework.

The Framework, Causal and Cocompact Structure of Spacetime ; We introduce a canonical, compact topology, which we call weakly causal, naturally generated by the causal site of J. D. Christensen and L. Crane, a pointless algebraic structure motivated by certain problems of quantum gravity. We show that for every fourdimensional globally hyperbolic Lorentzian manifold there exists an associated causal site, whose weakly causal topology is cocompact with respect to the manifold topology and vice versa. Thus, the causal site has the full information about the topology of spacetime, represented by the Lorentzian manifold. In addition, we show that there exist also nonLorentzian causal sites whose causal relation is not a continuous poset and so the weakly causal topology and its de Groot dual extends the usual manifold topology of spacetime beyond topologies generated by the traditional, smooth model. As a source of inspiration in topologizing the studied causal structures, we use some methods and constructions of general topology and formal concept analysis.

A Mixture of Coalesced Generalized Hyperbolic Distributions ; A mixture of multiple scaled generalized hyperbolic distributions MMSGHDs is introduced. Then, a coalesced generalized hyperbolic distribution CGHD is developed by joining a generalized hyperbolic distribution with a multiple scaled generalized hyperbolic distribution. After detailing the development of the MMSGHDs, which arises via implementation of a multidimensional weight function, the density of the mixture of CGHDs is developed. A parameter estimation scheme is developed using the everexpanding class of MM algorithms and the Bayesian information criterion is used for model selection. The issue of cluster convexity is examined and a special case of the MMSGHDs is developed that is guaranteed to have convex clusters. These approaches are illustrated and compared using simulated and real data. The identifiability of the MMSGHDs and the mixture of CGHDs is discussed in an appendix.

Integrable Motion of Curves in SelfConsistent Potentials Relation to Spin Systems and Soliton Equations ; Motion of curves and surfaces in R3 lead to nonlinear evolution equations which are often integrable. They are also intimately connected to the dynamics of spin chains in the continuum limit and integrable soliton systems through geometric and gauge symmetric connectionsequivalence. Here we point out the fact that a more general situation in which the curves evolve in the presence of additional self consistent vector potentials can lead to interesting generalized spin systems with self consistent potentials or soliton equations with self consistent potentials. We obtain the general form of the evolution equations of underlying curves and report specific examples of generalized spin chains and soliton equations. These include principal chiral model and various Myrzakulov spin equations in 11 dimensions and their geometrically equivalent generalized nonlinear Schrodinger NLS family of equations, including HirotaMaxwellBloch equations, all in the presence of self consistent potential fields. The associated gauge equivalent Lax pairs are also presented to confirm their integrability.

Extended generalized geometry and a DBItype effective action for branes ending on branes ; Starting from the usual bosonic membrane action, we develop the geometry suitable for the description of pbrane backgrounds. Using the tools of generalized geometry we derive the generalization of string openclosed relations. NambuPoisson structures are used to generalize the concept of semiclassical noncommutativity of Dbranes governed by Poisson tensor. We naturally describe the correspondence of recently proposed commutative and noncommutative versions of an effective action for pbranes ending on a p'brane. We calculate the power series expansion of the action in background independent gauge. Leading terms in the double scaling limit are given by a generalization of a semiclassical matrix model.

The Capacity Region of the SourceType Model for Secret Key and Private Key Generation ; The problem of simultaneously generating a secret key SK and private key PK pair among three terminals via public discussion is investigated, in which each terminal observes a component of correlated sources. All three terminals are required to generate a common secret key concealed from an eavesdropper that has access to public discussion, while two designated terminals are required to generate an extra private key concealed from both the eavesdropper and the remaining terminal. An outer bound on the SKPK capacity region was established in 1, and was shown to be achievable for one case. In this paper, achievable schemes are designed to achieve the outer bound for the remaining two cases, and hence the SKPK capacity region is established in general. The main technique lies in the novel design of a random binningjoint decoding scheme that achieves the existing outer bound.

Unifying framework for scalartensor theories of gravity ; A general framework for effective theories propagating two tensor and one scalar degrees of freedom is investigated. Geometrically, it describes dynamical foliation of spacelike hypersurfaces coupled to a general background, in which the scalar mode encodes the fluctuation of the hypersurfaces. Within this framework, various models in the literatureincluding kessence, Horndeski theory, the effective field theory of inflation, ghost condensate as well as the Hovrava gravityget unified. Our framework generalizes the Horndeski theory in the sense that, it propagates the correct number of degrees of freedom, although the equations of motion are generally higher order. We also identify new operators beyond the Horndeski theory, which yield second order equations of motion for linear perturbations around an a FriedmannRobertsonWalker background.

Generalized cable formalism to calculate the magnetic field of single neurons and neuronal populations ; Neurons generate magnetic fields which can be recorded with macroscopic techniques such as magnetoencephalography. The theory that accounts for the genesis of neuronal magnetic fields involves dendritic cable structures in homogeneous resistive extracellular media. Here, we generalize this model by considering dendritic cables in extracellular media with arbitrarily complex electric properties. This method is based on a multiscale meanfield theory where the neuron is considered in interaction with a mean extracellular medium characterized by a specific impedance. We first show that, as expected, the generalized cable equation and the standard cable generate magnetic fields that mostly depend on the axial current in the cable, with a moderate contribution of extracellular currents. Less expected, we also show that the nature of the extracellular and intracellular media influence the axial current, and thus also influence neuronal magnetic fields. We illustrate these properties by numerical simulations and suggest experiments to test these findings.

A Multiperiod OPF Model Under Renewable Generation Uncertainty and Demand Side Flexibility ; Renewable energy sources such as wind and solar have received much attention in recent years and large amount of renewable generation is being integrated to the electricity networks. A fundamental challenge in power system operation is to handle the intermittent nature of the renewable generation. In this paper we present a stochastic programming approach to solve a multiperiod optimal power flow problem under renewable generation uncertainty. The proposed approach consists of two stages. In the first stage operating points for conventional power plants are determined. Second stage realizes the generation from renewable resources and optimally accommodates it by relying on demandside flexibility. The benefits from its application are demonstrated and discussed on a 4bus and a 39bus systems. Numerical results show that with limited flexibility on the demandside substantial benefits in terms of potential additional redispatch costs can be achieved. The scaling properties of the approach are finally analysed based on standard IEEE test cases upto 300 buses, allowing to underlined its computational efficiency.

Generating kindependent variables in constant time ; The generation of pseudorandom elements over finite fields is fundamental to the time, space and randomness complexity of randomized algorithms and data structures. We consider the problem of generating kindependent random values over a finite field mathbbF in a word RAM model equipped with constant time addition and multiplication in mathbbF, and present the first nontrivial construction of a generator that outputs each value in constant time, not dependent on k. Our generator has period length mathbbF,mboxpoly log k and uses k,mboxpolylog k log mathbbF bits of space, which is optimal up to a mboxpoly log k factor. We are able to bypass Siegel's lower bound on the timespace tradeoff for kindependent functions by a restriction to sequential evaluation.

Generation bidding game with flexible demand ; For a simple model of priceresponsive demand, we consider a deregulated electricity marketplace wherein the grid ISO, retailerdistributor accepts bids perunit supply from generators simplified herein neither to consider startuprampup expenses nor dayahead or shorterterm load following which are then averaged by supply allocations via an economic dispatch to a common clearing price borne by customers irrespective of variations in transmissiondistribution or generation prices, i.e., the ISO does not compensate generators based on their marginal costs. Rather, the ISO provides sufficient information for generators to sensibly adjust their bids. Notwithstanding our idealizations, the dispatch dynamics are complex. For a simple benchmark power system, we find a pricesymmetric Nash equilibrium through numerical experiments.

Spin Generation Via Bulk Spin Current in Three Dimensional Topological Insulators ; To date, spin generation in threedimensional topological insulators is primarily modeled as a singlesurface phenomenon, attributed to the momentumspin locking on each individual surface. In this article we propose a mechanism of spin generation where the role of the insulating yet topologically nontrivial bulk becomes explicit an external electric field creates a transverse pure spin current through the bulk of a threedimensional topological insulator, which transports spins between the top and bottom surfaces. Under sufficiently high surface disorder, the spin relaxation time can be extended via the DyakonovPerel mechanism. Consequently both the spin generation efficiency and surface conductivity are largely enhanced. Numerical simulation confirms that this spin generation mechanism originates from the unique topological connection of the top and bottom surfaces and is absent in other two dimensional systems such as graphene, even though they possess a similar Dirac conetype dispersion.

Broadband, stable and highly coherent supercontinuum generation at telecommunication wavelengths in an hydrogenated amorphous silicon waveguide ; Hydrogenated amorphous silicon aSiH has recently been recognized as a highly nonlinear CMOS compatible photonic platform. We experimentally demonstrate the generation of a supercontinuum SC spanning over 500 nm in aSiH photonic wire waveguide at telecommunication wavelengths using femtosecond input pulse with energy lower than 5 pJ. Numerical modeling of pulse propagation in the waveguide, based on the experimentally characterized dispersion profile, shows that the supercontinuum is the result of soliton fission and dispersive wave generation. It is demonstrated that the SC is highly coherent and that the waveguides do not suffer from material degradation under femtosecond pulse illumination. Finally, a direct comparison of SC generation in cSi and aSiH waveguides confirms the higher performances of aSiH over cSi for broadband low power SC generation at telecommunication wavelengths.

Multilocality and fusion rules on the generalized structure functions in twodimensional and threedimensional NavierStokes turbulence ; Using the fusion rules hypothesis for threedimensional and twodimensional NavierStokes turbulence, we generalize a previous nonperturbative locality proof to multiple applications of the nonlinear interactions operator on generalized structure functions of velocity differences. We shall call this generalization of nonperturbative locality to multiple applications of the nonlinear interactions operator multilocality. The resulting crossterms pose a new challenge requiring a new argument and the introduction of a new fusion rule that takes advantage of rotational symmetry. Our main result is that the fusion rules hypothesis implies both locality and multilocality in both the IR and UV limits for the downscale energy cascade of threedimensional NavierStokes turbulence and the downscale enstrophy cascade and inverse energy cascade of twodimensional NavierStokes turbulence. We stress that these claims relate to nonperturbative locality of generalized structure functions on all orders, and not the term by term perturbative locality of diagrammatic theories or closure models that involve only twopoint correlation and response functions.

The role of currents distribution in general relativistic equilibria of magnetized neutron stars ; Magnetic fields play a critical role in the phenomenology of neutron stars. There is virtually no observable aspect which is not governed by them. Despite this, only recently efforts have been done to model magnetic fields in the correct general relativistic regime, characteristic of these compact objects. In this work we present, for the first time a comprehensive and detailed parameter study, in general relativity, of the role that the current distribution, and the related magnetic field structure, have in determining the precise structure of neutron stars. In particular, we show how the presence of localized currents can modify the field strength at the stellar surface, and we look for general trends, both in terms of energetic properties, and magnetic field configurations. Here we verify that, among other things, for a large class of different current distributions the resulting magnetic configurations are always dominated by the poloidal component of the current.

A Generalization of the SpaceFractional Poisson Process and its Connection to some Levy Processes ; This paper introduces a generalization of the socalled spacefractional Poisson process by extending the difference operator acting on state space present in the associated differencedifferential equations to a much more general form. It turns out that this generalization can be put in relation to a specific subordination of a homogeneous Poisson process by means of a subordinator for which it is possible to express the characterizing L'evy measure explicitly. Moreover, the law of this subordinator solves a onesided firstorder differential equation in which a particular convolutiontype integral operator appears, called Prabhakar derivative. In the last section of the paper, a similar model is introduced in which the Prabhakar derivative also acts in time. In this case, too, the probability generating function of the corresponding process and the probability distribution are determined.

Wishart Generator Distribution ; The Wishart distribution and its generalizations are among the most prominent probability distributions in multivariate statistical analysis, arising naturally in applied research and as a basis for theoretical models. In this paper, we generalize the Wishart distribution utilizing a different approach that leads to the Wishart generator distribution with the Wishart distribution as a special case. It is not restricted, however some special cases are exhibited. Important statistical characteristics of the Wishart generator distribution are derived from the matrix theory viewpoint. Estimation is also touched upon as a guide for further research from the classical approach as well as from the Bayesian paradigm. The paper is concluded by giving applications of two special cases of this distribution in calculating the product of beta functions and astronomy.

Symmetries of general nonMarkovian Gaussian diffusive unravelings ; By using a condition of average trace preservation we derive a general class of nonMarkovian Gaussian diffusive unravelings L. Diosi and L. Ferialdi, Phys. Rev. Lett. textbf113, 200403 2014, here valid for arbitrary nonHermitian system operators and noise correlations. The conditions under which the generalized stochastic Schrodinger equation has the same symmetry properties invariance under unitary changes of operator base than a microscopic systembath Hamiltonian dynamics are determined. While the standard quantum diffusion model standard noise correlations always share the same invariance symmetry, the generalized stochastic dynamics can be mapped with an arbitrary bosonic environment only if some specific correlation constraints are fulfilled. These features are analyzed for different nonMarkovian unravelings equivalent in average. Results based on quantum measurement theory that lead to specific cases of the generalized dynamics J. Gambetta and H. M. Wiseman, Phys. Rev. A textbf66, 012108 2002 are studied from the perspective of the present analysis.

Fractionalorder Generalized Principle of SelfSupport FOG PSS in Control Systems Design ; This paper reviews research that studies the principle of selfsupport PSS in some control systems and proposes a fractionalorder generalized PSS framework for the first time. The existing PSS approach focuses on practical tracking problem of integerorder systems including robotic dynamics, high precision linear motor system, multiaxis high precision positioning system with unmeasurable variables, imprecise sensor information, uncertain parameters and external disturbances. More generally, by formulating the fractional PSS concept as a new generalized framework, we will focus in the possible fields on the fractionalorder control problems such as practical tracking, lambdatracking, etc. of robot systems, multiple mobile agents, discrete dynamical systems, time delay systems and other uncertain nonlinear systems. Finally, the practical tracking of a firstorder uncertain model of automobile is considered as a simple example to demonstrate the efficiency of the fractionalorder generalized principle of selfsupport FOGPSS control strategy.

Special function identities from superelliptic Kummer varieties ; We prove that the factorization of Appell's generalized hypergeometric series satisfying the socalled quadric property into a product of two Gauss' hypergeometric functions has a geometric origin we first construct a generalized Kummer variety as minimal nonsingular model for a productquotient surface with only rational double points from a pair of superelliptic curves of genus 2r1 with r in mathbbN. We then show that this generalized Kummer variety is equipped with two fibrations with fibers of genus 2r1. When periods of a holomorphic twoform over carefully crafted transcendental twocycles on the generalized Kummer variety are evaluated using either of the two fibrations, the answer must be independent of the fibration and the aforementioned family of special function identities is obtained. This family of identities can be seen as a multivariate generalization of Clausen's Formula. Interestingly, this paper's finding bridges Ernst Kummer's two independent lines of research, algebraic transformations for the Gauss' hypergeometric function and nodal surfaces of degree four in mathbbP3.

Direct generation of optical frequency combs in 2 nonlinear cavities ; Quadratic nonlinear processes are currently exploited for frequency comb transfer and extension from the visible and near infrared regions to other spectral ranges where direct comb generation cannot be accomplished. However, frequency comb generation has been directly observed in continuouslypumped quadratic nonlinear crystals placed inside an optical cavity. At the same time, an introductory theoretical description of the phenomenon has been provided, showing a remarkable analogy with the dynamics of thirdorder Kerr microresonators. Here, we give an overview of our recent work on chi2 frequency comb generation. Furthermore, we generalize the preliminary threewave spectral model to a manymode comb and present a stability analysis of different cavity field regimes. Although at a very early stage, our work lays the groundwork for a novel class of highly efficient and versatile frequency comb synthesizers based on secondorder nonlinear materials.

Joint Investment and Operation of Microgrid ; In this paper, we propose a theoretical framework for the joint optimization of investment and operation of a microgrid, taking the impact of energy storage, renewable energy integration, and demand response into consideration. We first study the renewable energy generations in Hong kong, and identify the potential benefit of mixed deployment of solar and wind energy generations. Then we model the joint investment and operation as a twoperiod stochastic programming program. In period1, the microgrid operator makes the optimal investment decisions on the capacities of solar power generation, wind power generation, and energy storage. In period2, the operator coordinates the power supply and demand in the microgrid to minimize the operating cost. We design a decentralized algorithm for computing the optimal pricing and power consumption in period2, based on which we solve the optimal investment problem in period1. We also study the impact of prediction error of renewable energy generation on the portfolio investment using robust optimization framework. Using realistic meteorological data obtained from the Hong Kong observatory, we numerically characterize the optimal portfolio investment decisions, optimal dayahead pricing and power scheduling, and demonstrate the advantage of using mixed renewable energy and demand response in terms of reducing investment cost.

Generalized Higher Gauge Theory ; We study a generalization of higher gauge theory which makes use of generalized geometry and seems to be closely related to double field theory. The local kinematical data of this theory is captured by morphisms of graded manifolds between the canonical exact Courant Lie 2algebroid TMoplus TM over some manifold M and a semistrict gauge Lie 2algebra. We discuss generalized curvatures and their infinitesimal gauge transformations. Finite gauge transformation as well as global kinematical data are then obtained from principal 2bundles over 2spaces. As dynamical principle, we consider first the canonical ChernSimons action for such a gauge theory. We then show that a previously proposed 3Lie algebra model for the sixdimensional 2,0 theory is very naturally interpreted as a generalized higher gauge theory.

SecretKey Generation Using Compound Sources and OneWay Public Communication ; In the classical SecretKey generation model, Common Randomness is generated by two terminals based on the observation of correlated components of a common source, while keeping it secret from a nonlegitimate observer. It is assumed that the statistics of the source are known to all participants. In this work, the SecretKey generation based on a compound source is studied where the realization of the source statistic is unknown. The protocol should guarantee the security and reliability of the generated SecretKey, simultaneously for all possible realizations of the compound source. A singleletter lowerbound of the SecretKey capacity for a finite compound source is derived as a function of the public communication rate constraint. A multiletter capacity formula is further computed for a finite compound source for the case in which the public communication is unconstrained. Finally a singleletter capacity formula is derived for a degraded compound source with an arbitrary set of source states and a finite set of marginal states.

Generic appearance of objective results in quantum measurements ; Measurement is of central interest in quantum mechanics as it provides the link between the quantum world and the world of everyday experience. One of the features of the latter is its robust, objective character, contrasting the delicate nature of quantum systems. Here we analyze in a completely modelindependent way the celebrated von Neumann measurement process, using recent techniques of information flow, studied in open quantum systems. We show the generic appearance of objective results in quantum measurements, provided we macroscopically coarsegrain the measuring apparatus and wait long enough. To study genericity, we employ the widelyused Gaussian Unitary Ensemble of random matrices and the Hoeffding inequality. We derive generic objectivization timescales, given solely by the interaction strength and the systems' dimensions. Our results are manifestly universal and are a generic property of von Neumann measurements.

Dynamical Gravitational Coupling as a Modified Theory of General Relativity ; A modified theory of general relativity is proposed, where the gravitational constant is replaced by a dynamical variable in spacetime. The dynamics of the gravitational coupling is described by a family of parametrized null geodesics, implying that the gravitational coupling at a spacetime point is determined by solving transport equations along all null geodesics through this point. General relativity with dynamical gravitational coupling DGC is introduced. We motivate DGC from general considerations and explain how it arises in the context of causal fermion systems. The underlying physical idea is that the gravitational coupling is determined by microscopic structures on the Planck scale which propagate with the speed of light. In order to clarify the mathematical structure, we analyze the conformal behavior and prove local existence and uniqueness of the time evolution. The differences to Einstein's theory are worked out in the examples of the FriedmannRobertsonWalker model and the spherically symmetric collapse of a shell of matter. Potential implications for the problem of dark matter and for inflation are discussed. It is shown that the effects in the solar system are too small for being observable in presentday experiments.

Lowenergy Spin Dynamics of the Honeycomb Spin Liquid Beyond the Kitaev Limit ; We investigate the generic features of the low energy dynamical spin structure factor of the Kitaev honeycomb quantum spin liquid perturbed away from its exact soluble limit by generic symmetryallowed exchange couplings. We find that the spin gap persists in the KitaevHeisenberg model, but generally vanishes provided more generic symmetryallowed interactions exist. We formulate the generic expansion of the spin operator in terms of fractionalized Majorana fermion operators according to the symmetry enriched topological order of the Kitaev spin liquid, described by its projective symmetry group. The dynamical spin structure factor displays powerlaw scaling bounded by Dirac cones in the vicinity of the Gamma, K and K' points of the Brillouin zone, rather than the spin gap found for the exactly soluble point.

A Novel Approach to Dropped Pronoun Translation ; Dropped Pronouns DP in which pronouns are frequently dropped in the source language but should be retained in the target language are challenge in machine translation. In response to this problem, we propose a semisupervised approach to recall possibly missing pronouns in the translation. Firstly, we build training data for DP generation in which the DPs are automatically labelled according to the alignment information from a parallel corpus. Secondly, we build a deep learningbased DP generator for input sentences in decoding when no corresponding references exist. More specifically, the generation is twophase 1 DP position detection, which is modeled as a sequential labelling task with recurrent neural networks; and 2 DP prediction, which employs a multilayer perceptron with rich features. Finally, we integrate the above outputs into our translation system to recall missing pronouns by both extracting rules from the DPlabelled training data and translating the DPgenerated input sentences. Experimental results show that our approach achieves a significant improvement of 1.58 BLEU points in translation performance with 66 Fscore for DP generation accuracy.

Generic instabilities of nonsingular cosmologies in Horndeski theory a nogo theorem ; The null energy condition can be violated stably in generalized Galileon theories, which gives rise to the possibilities of healthy nonsingular cosmologies. However, it has been reported that in many cases cosmological solutions are plagued with instabilities or have some pathologies somewhere in the whole history of the universe. Recently, this was shown to be generically true in a certain subclass of the Horndeski theory. In this short paper, we extend this nogo argument to the full Horndeski theory, and show that nonsingular models with flat spatial sections in general suffer either from gradient instabilities or some kind of pathology in the tensor sector. This implies that one must go beyond the Horndeski theory to implement healthy nonsingular cosmologies.

LCrowdV Generating Labeled Videos for Simulationbased Crowd Behavior Learning ; We present a novel procedural framework to generate an arbitrary number of labeled crowd videos LCrowdV. The resulting crowd video datasets are used to design accurate algorithms or training models for crowded scene understanding. Our overall approach is composed of two components a procedural simulation framework for generating crowd movements and behaviors, and a procedural rendering framework to generate different videos or images. Each video or image is automatically labeled based on the environment, number of pedestrians, density, behavior, flow, lighting conditions, viewpoint, noise, etc. Furthermore, we can increase the realism by combining syntheticallygenerated behaviors with realworld background videos. We demonstrate the benefits of LCrowdV over prior lableled crowd datasets by improving the accuracy of pedestrian detection and crowd behavior classification algorithms. LCrowdV would be released on the WWW.

PostNewtonian parameter for multiscalartensor gravity with a general potential ; We compute the parametrized postNewtonian parameter gamma in the case of a static point source for multiscalartensor gravity with completely general nonderivative couplings and potential in the Jordan frame. Similarly to the single massive field case gamma depends exponentially on the distance from the source and is determined by the length of a vector of nonminimal coupling in the space of scalar fields and its orientation relative to the mass eigenvectors. Using data from the Cassini tracking experiment, we estimate bounds on a general theory with two scalar fields. Our formalism can be utilized for a wide range of models, which we illustrate by applying it to nonminimally coupled Higgs SU2 doublet, general hybrid metricPalatini gravity, linear Box1 and quadratic Box2 nonlocal gravity.

Degravitation, Orbital Dynamics and the Effective Barycentre ; In this article we present a particular theory of gravity in which Einstein's field equations are modified by promoting Newton's constant G to a covariant differential operator GLambdaBoxg. The general idea was obviously outlined for the first time in 1316 and originates from the quest of finding a mechanism that is able to degravitate the vacuum energy on cosmological scales. We suggest in this manuscript a precise covariant coupling model which acts like a highpass filter with a macroscopic distance filter scale sqrtLambda. In the context of this specific theory of gravity we review some cosmological aspects before we briefly recall the effective relaxed Einstein equations outlined for the first time in 1. We present a general procedure to determine the gravitational potentials for a far away wave zone field point. Moreover we work out the modified orbital dynamics of a binarysystem as well as the effective 1.5 postNewtonian barycentre for a generic nbody system. We notice that it is always possible to recover the corresponding general relativistic results in the limit of vanishing nonlocal modification parameters.

Energybased Generative Adversarial Network ; We introduce the Energybased Generative Adversarial Network model EBGAN which views the discriminator as an energy function that attributes low energies to the regions near the data manifold and higher energies to other regions. Similar to the probabilistic GANs, a generator is seen as being trained to produce contrastive samples with minimal energies, while the discriminator is trained to assign high energies to these generated samples. Viewing the discriminator as an energy function allows to use a wide variety of architectures and loss functionals in addition to the usual binary classifier with logistic output. Among them, we show one instantiation of EBGAN framework as using an autoencoder architecture, with the energy being the reconstruction error, in place of the discriminator. We show that this form of EBGAN exhibits more stable behavior than regular GANs during training. We also show that a singlescale architecture can be trained to generate highresolution images.

Mode Regularized Generative Adversarial Networks ; Although Generative Adversarial Networks achieve stateoftheart results on a variety of generative tasks, they are regarded as highly unstable and prone to miss modes. We argue that these bad behaviors of GANs are due to the very particular functional shape of the trained discriminators in high dimensional spaces, which can easily make training stuck or push probability mass in the wrong direction, towards that of higher concentration than that of the data generating distribution. We introduce several ways of regularizing the objective, which can dramatically stabilize the training of GAN models. We also show that our regularizers can help the fair distribution of probability mass across the modes of the data generating distribution, during the early phases of training and thus providing a unified solution to the missing modes problem.

AffleckDine Leptogenesis with Varying PecceiQuinn Scale ; The AffleckDine leptogenesis scenario along the LHu flat direction is reconsidered. It is known that successful AffleckDine leptogenesis requires that the lightest neutrino mass is extremely small. This situation can be significantly relaxed if the neutrino mass in the early universe is different from the present one. We consider a supersymmetric DineFischlerSrednickiZhitnitsky DFSZ type model, which provides a solution to the strong CP problem and generates a SUSY muterm and righthanded neutrino masses. If the PQ scale during lepton number generation is much larger than the present value, leptogenesis is very efficient so that enough baryon number can be generated without introducing a hierarchically small neutrino mass. The final baryon asymmetry is related to the muterm, and hence linked to the level of electroweak finetuning. We also show the PQ breaking scalar dynamics that keeps a large PQ breaking scale during inflation and lepton number generation. The muterm generating superpotential plays an important role for preserving the lepton asymmetry during saxion oscillation. In this scenario, the axion isocurvature perturbation is naturally suppressed.

Imposing higherlevel Structure in Polyphonic Music Generation using Convolutional Restricted Boltzmann Machines and Constraints ; We introduce a method for imposing higherlevel structure on generated, polyphonic music. A Convolutional Restricted Boltzmann Machine CRBM as a generative model is combined with gradient descent constraint optimisation to provide further control over the generation process. Among other things, this allows for the use of a template piece, from which some structural properties can be extracted, and transferred as constraints to the newly generated material. The sampling process is guided with Simulated Annealing to avoid local optima, and to find solutions that both satisfy the constraints, and are relatively stable with respect to the CRBM. Results show that with this approach it is possible to control the higherlevel selfsimilarity structure, the meter, and the tonal properties of the resulting musical piece, while preserving its local musical coherence.

A Note on One Less Known Class of Generated Residual Implications ; This paper builds on our contribution Havlena and Hlinena, 2016 which studied modelling of the conjunction in human language. We have discussed three different ways of constructing conjunction. We have dealt with generated tnorms, generated means and Choquet integral. In this paper we construct the residual operators based on the above conjunctions. The only operator based on a tnorm is an implication. We show that this implication belongs to the class of generated implications IgN which was introduced in Smutna, 1999 and studied in Biba and Hlinena, 2012. We study its properties. More, we investigate this class of generated implications. Some important properties, including relations between some classes of implications, are given.

General Relativity a la string a progress report ; Preliminary results on a canonical formulation of general relativity based on an analogy with the string model of elementary particles are presented. Rather than the metric components, the basic fields of the formalism are taken to be the functions describing the embedding of four dimensional spacetime in a ten or possibly higher dimensional manifold. So far, the main drawback of the formalism is that the generator of normal deformations fourth constraint cannot be written down in closed form. The present approach is compared and contrasted with the usual one and with the canonical description of the relativistic string.

Unsupervised PixelLevel Domain Adaptation with Generative Adversarial Networks ; Collecting wellannotated image datasets to train modern machine learning algorithms is prohibitively expensive for many tasks. One appealing alternative is rendering synthetic data where groundtruth annotations are generated automatically. Unfortunately, models trained purely on rendered images often fail to generalize to real images. To address this shortcoming, prior work introduced unsupervised domain adaptation algorithms that attempt to map representations between the two domains or learn to extract features that are domaininvariant. In this work, we present a new approach that learns, in an unsupervised manner, a transformation in the pixel space from one domain to the other. Our generative adversarial network GANbased method adapts sourcedomain images to appear as if drawn from the target domain. Our approach not only produces plausible samples, but also outperforms the stateoftheart on a number of unsupervised domain adaptation scenarios by large margins. Finally, we demonstrate that the adaptation process generalizes to object classes unseen during training.

Intense keV isolated attosecond pulse generation by orthogonally polarized multicycle midinfrared twocolor laser field ; We theoretically investigate the generation of intense keV attosecond pulses in an orthogonally polarized multicycle midinfrared twocolor laser field. It is demonstrated that multiple continuumlike humps, which have a spectral width of about twenty orders of harmonics and an intensity of about one order higher than adjacent normal harmonic peaks, are generated under proper twocolor delays, owing to the reduction of the number of electronion recollisions and suppression of interhalfcycle interference effect of multiple electron trajectories when the long wavelength midinfrared driving field is used. Using the semiclassical trajectory model, we have revealed the twodimensional manipulation of the electronion recollision process, which agrees well with the time frequency analysis. By filtering these humps, intense isolated attosecond pulses are directly generated without any phase compensation. Our proposal provides a simple technique to generate intense isolated attosecond pulses with various central photon energies covering the multikeV spectral regime by using multicycle driving pulses with high pump energy in experiment.

Generic Phase Portrait Analysis of the Finitetime Singularities and Generalized Teleparallel Gravity ; We analyze the common four types of the finitetime singularities using a generic framework of the phase portrait geometric approach. This technique requires that the Friedmann system to be written as a one dimensional autonomous system. We employ a scale factor that has been used widely in literature to realize the four finitetime singularity types, then we show a detailed discussion for each case showing possible novel models. Moreover, we show how different singularity types can play essential roles in different cosmological scenarios. Among several modified gravity theories, we show that the f T cosmology is in comfort with the phase portrait analysis, since the field equations include Hubble derivatives only up to first order. Therefore, we reconstruct the f T theory which generates these phase portraits. We also perform a complementary analysis using the effective equation of state. Furthermore, we investigate the role of the torsion fluid in realizing the cosmic singularities.

Soft Label MemorizationGeneralization for Natural Language Inference ; Often when multiple labels are obtained for a training example it is assumed that there is an element of noise that must be accounted for. It has been shown that this disagreement can be considered signal instead of noise. In this work we investigate using soft labels for training data to improve generalization in machine learning models. However, using soft labels for training Deep Neural Networks DNNs is not practical due to the costs involved in obtaining multiple labels for large data sets. We propose soft label memorizationgeneralization SLMG, a finetuning approach to using soft labels for training DNNs. We assume that differences in labels provided by human annotators represent ambiguity about the true label instead of noise. Experiments with SLMG demonstrate improved generalization performance on the Natural Language Inference NLI task. Our experiments show that by injecting a small percentage of soft label training data 0.03 of training set size we can improve generalization performance over several baselines.

Sampling Variations of Lead Sheets ; Machinelearning techniques have been recently used with spectacular results to generate artefacts such as music or text. However, these techniques are still unable to capture and generate artefacts that are convincingly structured. In this paper we present an approach to generate structured musical sequences. We introduce a mechanism for sampling efficiently variations of musical sequences. Given a input sequence and a statistical model, this mechanism samples a set of sequences whose distance to the input sequence is approximately within specified bounds. This mechanism is implemented as an extension of belief propagation, and uses local fields to bias the generation. We show experimentally that sampled sequences are indeed closely correlated to the standard musical similarity measure defined by Mongeau and Sankoff. We then show how this mechanism can used to implement composition strategies that enforce arbitrary structure on a musical lead sheet generation problem.

SurfNet Generating 3D shape surfaces using deep residual networks ; 3D shape models are naturally parameterized using vertices and faces, ie, composed of polygons forming a surface. However, current 3D learning paradigms for predictive and generative tasks using convolutional neural networks focus on a voxelized representation of the object. Lifting convolution operators from the traditional 2D to 3D results in high computational overhead with little additional benefit as most of the geometry information is contained on the surface boundary. Here we study the problem of directly generating the 3D shape surface of rigid and nonrigid shapes using deep convolutional neural networks. We develop a procedure to create consistent geometry images' representing the shape surface of a category of 3D objects. We then use this consistent representation for categoryspecific shape surface generation from a parametric representation or an image by developing novel extensions of deep residual networks for the task of geometry image generation. Our experiments indicate that our network learns a meaningful representation of shape surfaces allowing it to interpolate between shape orientations and poses, invent new shape surfaces and reconstruct 3D shape surfaces from previously unseen images.

Covariant Generalized Holographic Dark Energy and Accelerating Universe ; We proposed the generalized holographic dark energy model where infrared cutoff is identified with the combination of the FRW universe parameters the Hubble rate, particle and future horizons, cosmological constant, the universe lifetime if finite and their derivatives. It is demonstrated that with the corresponding choice of the cutoff one can map such holographic dark energy to modified gravity or gravity with general fluid. Explicitly, FR gravity and general perfect fluid are worked out in detail and corresponding infrared cutoff is found. Using this correspondence, we get realistic inflation or viable dark energy or unified inflationarydark energy universe in terms of covariant holographic dark energy.

Generative Adversarial Residual Pairwise Networks for One Shot Learning ; Deep neural networks achieve unprecedented performance levels over many tasks and scale well with large quantities of data, but performance in the lowdata regime and tasks like one shot learning still lags behind. While recent work suggests many hypotheses from better optimization to more complicated network structures, in this work we hypothesize that having a learnable and more expressive similarity objective is an essential missing component. Towards overcoming that, we propose a network design inspired by deep residual networks that allows the efficient computation of this more expressive pairwise similarity objective. Further, we argue that regularization is key in learning with small amounts of data, and propose an additional generator network based on the Generative Adversarial Networks where the discriminator is our residual pairwise network. This provides a strong regularizer by leveraging the generated data samples. The proposed model can generate plausible variations of exemplars over unseen classes and outperforms strong discriminative baselines for few shot classification tasks. Notably, our residual pairwise network design outperforms previous stateoftheart on the challenging miniImagenet dataset for one shot learning by getting over 55 accuracy for the 5way classification task over unseen classes.

Weyl invariance for generalized supergravity backgrounds from the doubled formalism ; It has recently been shown that a set of the generalized type IIB supergravity equations follows from the requirement of kappa symmetry of the type IIB GreenSchwarz superstring theory defined on an arbitrary background. In this paper, we show that the whole bosonic part of the generalized type II supergravity equations can be reproduced from the Tduality covariant equations of motion of the double field theory by choosing a nonstandard solution of the strong constraint. Then, by using the doubled formalism, we show the Weyl invariance of the bosonic string sigma model on a generalized gravity background. According to the dualcoordinate dependence of the dilaton, the FradkinTseytlin term nicely removes the Weyl anomaly. This result seems likely to support that string theories can be consistently defined on arbitrary generalized supergravity backgrounds.

On the canonical structure of general relativity with a limiting curvature and its relation to loop quantum gravity ; Chamseddine and Mukhanov recently proposed a modified version of general relativity that implements the idea of a limiting curvature. In the spatially flat, homogeneous, and isotropic sector, their theory turns out to agree with the effective dynamics of the simplest version of loop quantum gravity if one identifies their limiting curvature with a multiple of the Planck curvature. At the same time, it extends to full general relativity without any symmetry assumptions and thus provides an ideal toy model for full loop quantum gravity in the form of a generally covariant effective action known to all orders. In this paper, we study the canonical structure of this theory and point out some interesting lessons for loop quantum gravity. We also highlight in detail how the two theories are connected in the spatially flat, homogeneous, and isotropic sector.

Generate To Adapt Aligning Domains using Generative Adversarial Networks ; Domain Adaptation is an actively researched problem in Computer Vision. In this work, we propose an approach that leverages unsupervised data to bring the source and target distributions closer in a learned joint feature space. We accomplish this by inducing a symbiotic relationship between the learned embedding and a generative adversarial network. This is in contrast to methods which use the adversarial framework for realistic data generation and retraining deep models with such data. We demonstrate the strength and generality of our approach by performing experiments on three different tasks with varying levels of difficulty 1 Digit classification MNIST, SVHN and USPS datasets 2 Object recognition using OFFICE dataset and 3 Domain adaptation from synthetic to real data. Our method achieves stateofthe art performance in most experimental settings and by far the only GANbased method that has been shown to work well across different datasets such as OFFICE and DIGITS.

Quasinormal modes as a distinguisher between general relativity and fR gravity ; QuasiNormal Modes QNM or ringdown phase of gravitational waves provide critical information about the structure of compact objects like Black Holes. Thus, QNMs can be a tool to test General Relativity GR and possible deviations from it. In the case of GR, it is known for a long time that a relation between two types of Black Hole perturbations scalar Zerilli and vector ReggeWheeler, leads to an equal share of emitted gravitational energy. With the direct detection of Gravitational waves, it is now natural to ask whether the same relation between scalar and vector perturbations holds for modified gravity theories If not, whether one can use this as a way to probe deviations from General Relativity. As a first step, we show explicitly that the above relation between ReggeWheeler and Zerilli breaks down for a general f R model, and hence the two perturbations do not share equal amounts of emitted gravitational energy. We discuss the implication of this imbalance on observations and the nohair conjecture.

Group invariance principles for causal generative models ; The postulate of independence of cause and mechanism ICM has recently led to several new causal discovery algorithms. The interpretation of independence and the way it is utilized, however, varies across these methods. Our aim in this paper is to propose a group theoretic framework for ICM to unify and generalize these approaches. In our setting, the causemechanism relationship is assessed by comparing it against a null hypothesis through the application of random generic group transformations. We show that the group theoretic view provides a very general tool to study the structure of data generating mechanisms with direct applications to machine learning.

Hairy blackhole solutions in generalized Proca theories ; We present a family of exact blackhole solutions on a static spherically symmetric background in secondorder generalized Proca theories with derivative vectorfield interactions coupled to gravity. We also derive nonexact solutions in powerlaw coupling models including vector Galileons and numerically show the existence of regular black holes with a primary hair associated with the longitudinal propagation. The intrinsic vectorfield derivative interactions generally give rise to a secondary hair induced by nontrivial field profiles. The deviation from General Relativity is most significant around the horizon and hence there is a golden opportunity for probing the Proca hair by the measurements of gravitational waves GWs in the regime of strong gravity.

Natural Language Generation for Spoken Dialogue System using RNN EncoderDecoder Networks ; Natural language generation NLG is a critical component in a spoken dialogue system. This paper presents a Recurrent Neural Network based EncoderDecoder architecture, in which an LSTMbased decoder is introduced to select, aggregate semantic elements produced by an attention mechanism over the input elements, and to produce the required utterances. The proposed generator can be jointly trained both sentence planning and surface realization to produce natural language sentences. The proposed model was extensively evaluated on four different NLG datasets. The experimental results showed that the proposed generators not only consistently outperform the previous methods across all the NLG domains but also show an ability to generalize from a new, unseen domain and learn from multidomain datasets.

A Novel Foamy Origin for Singlet Fermion Masses ; We show how masses for singlet fermions can be generated by interactions with a Dparticle model of spacetime foam inspired by brane theory. It has been shown previously by one of the authors N.E.M. such interactions may generate generate dynamically small masses for charged fermions via the recoils of Dparticle defects interacting with photons. In this work we consider the direct interactions of Dparticle with uncharged singlet fermions such as righthanded neutrinos. Quantum fluctuations of the lattice of Dparticles have massless vector spinone excitations that are analogues of phonons. These mediate forces between the singlet fermions, generating large dynamical masses that may be communicated to light neutrinos via the seesaw mechanism.

Relevance of Unsupervised Metrics in TaskOriented Dialogue for Evaluating Natural Language Generation ; Automated metrics such as BLEU are widely used in the machine translation literature. They have also been used recently in the dialogue community for evaluating dialogue response generation. However, previous work in dialogue response generation has shown that these metrics do not correlate strongly with human judgment in the non taskoriented dialogue setting. Taskoriented dialogue responses are expressed on narrower domains and exhibit lower diversity. It is thus reasonable to think that these automated metrics would correlate well with human judgment in the taskoriented setting where the generation task consists of translating dialogue acts into a sentence. We conduct an empirical study to confirm whether this is the case. Our findings indicate that these automated metrics have stronger correlation with human judgments in the taskoriented setting compared to what has been observed in the non taskoriented setting. We also observe that these metrics correlate even better for datasets which provide multiple ground truth reference sentences. In addition, we show that some of the currently available corpora for taskoriented language generation can be solved with simple models and advocate for more challenging datasets.

Studying Cascading Overload Failures under High Penetration of Wind Generation ; While power systems are reliable infrastructures, their complex interconnectivities allow for propagation of disturbances through cascading failures which causes blackouts. Meanwhile the ever increasing penetration level of renewable generation into power grids introduces a massive amount of uncertainty to the grid that might have a severe impact on grid vulnerability to overload cascading failures. There are numerous studies in the literature that focus on modeling cascading failures with different approaches. However, there is a need for studies that simulate cascading failure considering the uncertainty coming from high penetration of renewable generation. In this study, the impacts of wind generation in terms of its penetration and uncertainty levels on grid vulnerability to cascading overload failures are studied. The simulation results on IEEE 300 bus system show that uncertainty coming from wind energy have severe impact on grid vulnerability to cascading overload failures. Results also suggest that higher penetration levels of wind energy if not managed appropriately will add to this severity due to injection of higher uncertainties into the grid.

Warped Product Spacetimes ; Many classical results in relativity theory concerning spherically symmetric spacetimes have easy generalizations to warped product spacetimes, with a twodimensional Lorentzian base and arbitrary dimensional Riemannian fibers. We first give a systematic presentation of the main geometric constructions, with emphasis on the Kodama vector field and the Hawking energy; the construction is signature independent. This leads to proofs of general Birkhofftype theorems for warped product manifolds; our theorems in particular apply to situations where the warped product manifold is not necessarily Einstein, and thus can be applied to solutions with matter content in general relativity. Next we specialize to the Lorentzian case and study the propagation of null expansions under the assumption of the dominant energy condition. We prove several nonexistence results relating to the Yamabe class of the fibers, in the spirit of the blackhole topology theorem of HawkingGallowaySchoen. Finally we discuss the effect of the warped product ansatz on matter models. In particular we construct several cosmological solutions to the EinsteinEuler equations whose spatial geometry is generally not isotropic.

Data Sets Word Embeddings Learned from Tweets and General Data ; A word embedding is a lowdimensional, dense and real valued vector representation of a word. Word embeddings have been used in many NLP tasks. They are usually gener ated from a large text corpus. The embedding of a word cap tures both its syntactic and semantic aspects. Tweets are short, noisy and have unique lexical and semantic features that are different from other types of text. Therefore, it is necessary to have word embeddings learned specifically from tweets. In this paper, we present ten word embedding data sets. In addition to the data sets learned from just tweet data, we also built embedding sets from the general data and the combination of tweets with the general data. The general data consist of news articles, Wikipedia data and other web data. These ten embedding models were learned from about 400 million tweets and 7 billion words from the general text. In this paper, we also present two experiments demonstrating how to use the data sets in some NLP tasks, such as tweet sentiment analysis and tweet topic classification tasks.

Spincaloritronic signal generation in nondegenerate Si ; Spincaloritronic signal generation due to thermal spin injection and spin transport is demonstrated in a nondegenerate Si spin valve. The spindependent Seebeck effect is used for the spincaloritronic signal generation, and the thermal gradient of about 200 mK at an interface of Fe and Si enables generating a spin voltage of 8 muV at room temperature. A simple expansion of a conventional spin driftdiffusion model with taking into account the spindependent Seebeck effect shows semiconductor materials are quite potential for the spincaloritronic signal generation comparing with metallic materials, which can allow efficient heat recycling in semiconductor spin devices.

Limiting the Reconstruction Capability of Generative Neural Network using Negative Learning ; Generative models are widely used for unsupervised learning with various applications, including data compression and signal restoration. Training methods for such systems focus on the generality of the network given limited amount of training data. A less researched type of techniques concerns generation of only a single type of input. This is useful for applications such as constraint handling, noise reduction and anomaly detection. In this paper we present a technique to limit the generative capability of the network using negative learning. The proposed method searches the solution in the gradient direction for the desired input and in the opposite direction for the undesired input. One of the application can be anomaly detection where the undesired inputs are the anomalous data. In the results section we demonstrate the features of the algorithm using MNIST handwritten digit dataset and latter apply the technique to a realworld obstacle detection problem. The results clearly show that the proposed learning technique can significantly improve the performance for anomaly detection.

Steering Output Style and Topic in Neural Response Generation ; We propose simple and flexible training and decoding methods for influencing output style and topic in neural encoderdecoder based language generation. This capability is desirable in a variety of applications, including conversational systems, where successful agents need to produce language in a specific style and generate responses steered by a human puppeteer or external knowledge. We decompose the neural generation process into empirically easier subproblems a faithfulness model and a decoding method based on selectivesampling. We also describe training and sampling algorithms that bias the generation process with a specific language style restriction, or a topic restriction. Human evaluation results show that our proposed methods are able to restrict style and topic without degrading output quality in conversational tasks.

Fast Spherical Centroidal Voronoi Mesh Generation A Lloydpreconditioned LBFGS Method in Parallel ; Centroidal Voronoi tessellation CVTbased mesh generation is a very effective technique for creating highquality Voronoi meshes and their dual Delaunay triangulations that often play a crucial role in applications, including ocean and atmospheric simulations using finite volume schemes. In the next generation climate models, the spacing scales change dramatically across the whole sphere and require ultrahigh resolution and smooth transitions from coarse to fine grid regions. Thus fast and robust spherical CVT SCVT meshing algorithms become highly desirable. In this paper, we first propose a Lloydpreconditioned limitedmemory BFGS method for constructing SCVTs that is also applicable to the construction of CVTs of general domains. This method is then parallelized based on overlapping domain decomposition, enabling excellent scalability on distributed systems. Results of several computational experiments show that the new method could incur computational time costs one order of magnitude smaller compared with some existing methods for generating largescale highly variableresolution meshes, while also providing significantly improvements in mesh quality.

ANSAC Adaptive Nonminimal Sample and Consensus ; While RANSACbased methods are robust to incorrect image correspondences outliers, their hypothesis generators are not robust to correct image correspondences inliers with positional error noise. This slows down their convergence because hypotheses drawn from a minimal set of noisy inliers can deviate significantly from the optimal model. This work addresses this problem by introducing ANSAC, a RANSACbased estimator that accounts for noise by adaptively using more than the minimal number of correspondences required to generate a hypothesis. ANSAC estimates the inlier ratio the fraction of correct correspondences of several ranked subsets of candidate correspondences and generates hypotheses from them. Its hypothesisgeneration mechanism prioritizes the use of subsets with high inlier ratio to generate highquality hypotheses. ANSAC uses an early termination criterion that keeps track of the inlier ratio history and terminates when it has not changed significantly for a period of time. The experiments show that ANSAC finds good homography and fundamental matrix estimates in a few iterations, consistently outperforming stateoftheart methods.

Fastconverging Conditional Generative Adversarial Networks for Image Synthesis ; Building on top of the success of generative adversarial networks GANs, conditional GANs attempt to better direct the data generation process by conditioning with certain additional information. Inspired by the most recent ACGAN, in this paper we propose a fastconverging conditional GAN FCGAN. In addition to the realfake classifier used in vanilla GANs, our discriminator has an advanced auxiliary classifier which distinguishes each real class from an extra fake' class. The fake' class avoids mixing generated data with real data, which can potentially confuse the classification of real data as ACGAN does, and makes the advanced auxiliary classifier behave as another realfake classifier. As a result, FCGAN can accelerate the process of differentiation of all classes, thus boost the convergence speed. Experimental results on image synthesis demonstrate our model is competitive in the quality of images generated while achieving a faster convergence rate.

Implementation of True Random Number Generator based on DoubleScroll Attractor circuit with GST memristor emulator ; The cryptographic security provided by various techniques of random number generator RNG construction is one of the developing researches areas today. Among various types of RNG, the true random bit generator TRBG can be considered as the most unpredictable and most secured because its randomness seed is generated from chaotic sources. This paper proposes a design of TRBG model based on doublescroll attractors circuits with GST memristor. After implementation and simulation of the chaotic circuit with GST memristor emulator, the chaotic behavior of the output voltage and inductor current were received. Moreover, their dependence on the input voltage revealed the close to doublescroll form. The randomness generated from the proposed circuit was tested by receiving Fast Fourier Transform FFT and Lyapunov exponents of the output voltage.

Improving Image Captioning with Conditional Generative Adversarial Nets ; In this paper, we propose a novel conditionalgenerativeadversarialnetsbased image captioning framework as an extension of traditional reinforcementlearning RLbased encoderdecoder architecture. To deal with the inconsistent evaluation problem among different objective language metrics, we are motivated to design some discriminator networks to automatically and progressively determine whether generated caption is human described or machine generated. Two kinds of discriminator architectures CNN and RNNbased structures are introduced since each has its own advantages. The proposed algorithm is generic so that it can enhance any existing RLbased image captioning framework and we show that the conventional RL training method is just a special case of our approach. Empirically, we show consistent improvements over all language evaluation metrics for different stateoftheart image captioning models. In addition, the welltrained discriminators can also be viewed as objective image captioning evaluators

Generation of ringshaped optical vortices in dissipative media by inhomogeneous effective diffusion ; By means of systematic simulations we demonstrate generation of a variety of ringshaped optical vortices OVs from a twodimensional input with embedded vorticity, in a dissipative medium modeled by the cubicquintic complex GinzburgLandau equation with an inhomogeneous effective diffusion spatialfiltering term, which is anisotropic in the transverse plane and periodically modulated in the longitudinal direction. We show the generation of stable square and gearshaped OVs, as well as tilted ovalshaped vortex rings, and stringshaped bound states built of a central fundamental soliton and two vortex satellites, or of three fundamental solitons. Their shape can be adjusted by tuning the strength and modulation period of the inhomogeneous diffusion. Stability domains of the generated OVs are identified by varying the vorticity of the input and parameters of the inhomogeneous diffusion. The results suggest a method to generate new types of ringshaped OVs with applications to the work with structured light.

Stochastic Dynamics for Video Infilling ; In this paper, we introduce a stochastic dynamics video infilling SDVI framework to generate frames between long intervals in a video. Our task differs from video interpolation which aims to produce transitional frames for a short interval between every two frames and increase the temporal resolution. Our task, namely video infilling, however, aims to infill long intervals with plausible frame sequences. Our framework models the infilling as a constrained stochastic generation process and sequentially samples dynamics from the inferred distribution. SDVI consists of two parts 1 a bidirectional constraint propagation module to guarantee the spatialtemporal coherence among frames, 2 a stochastic sampling process to generate dynamics from the inferred distributions. Experimental results show that SDVI can generate clear frame sequences with varying contents. Moreover, motions in the generated sequence are realistic and able to transfer smoothly from the given start frame to the terminal frame. Our project site is httpsxharlie.github.ioprojectsprojectsitesSDVIvideoresults.html

Handwriting styles benchmarks and evaluation metrics ; Evaluating the style of handwriting generation is a challenging problem, since it is not well defined. It is a key component in order to develop in developing systems with more personalized experiences with humans. In this paper, we propose baseline benchmarks, in order to set anchors to estimate the relative quality of different handwriting style methods. This will be done using deep learning techniques, which have shown remarkable results in different machine learning tasks, learning classification, regression, and most relevant to our work, generating temporal sequences. We discuss the challenges associated with evaluating our methods, which is related to evaluation of generative models in general. We then propose evaluation metrics, which we find relevant to this problem, and we discuss how we evaluate the evaluation metrics. In this study, we use IRONOFF dataset. To the best of our knowledge, there is no work done before in generating handwriting either in terms of methodology or the performance metrics, our in exploring styles using this dataset.

C4Synth CrossCaption CycleConsistent TexttoImage Synthesis ; Generating an image from its description is a challenging task worth solving because of its numerous practical applications ranging from image editing to virtual reality. All existing methods use one single caption to generate a plausible image. A single caption by itself, can be limited, and may not be able to capture the variety of concepts and behavior that may be present in the image. We propose two deep generative models that generate an image by making use of multiple captions describing it. This is achieved by ensuring 'CrossCaption Cycle Consistency' between the multiple captions and the generated images. We report quantitative and qualitative results on the standard CaltechUCSD Birds CUB and Oxford102 Flowers datasets to validate the efficacy of the proposed approach.

SALSATEXT self attentive latent space based adversarial text generation ; Inspired by the success of self attention mechanism and Transformer architecture in sequence transduction and image generation applications, we propose novel self attentionbased architectures to improve the performance of adversarial latent code based schemes in text generation. Adversarial latent codebased text generation has recently gained a lot of attention due to their promising results. In this paper, we take a step to fortify the architectures used in these setups, specifically AAE and ARAE. We benchmark two latent codebased methods AAE and ARAE designed based on adversarial setups. In our experiments, the Google sentence compression dataset is utilized to compare our method with these methods using various objective and subjective measures. The experiments demonstrate the proposed self attentionbased models outperform the stateoftheart in adversarial codebased text generation.

The NIGENS General Sound Events Database ; Computational auditory scene analysis is gaining interest in the last years. Trailing behind the more mature field of speech recognition, it is particularly general sound event detection that is attracting increasing attention. Crucial for training and testing reasonable models is having available enough suitable data until recently, general sound event databases were hardly found. We release and present a database with 714 wav files containing isolated high quality sound events of 14 different types, plus 303 general' wav files of anything else but these 14 types. All sound events are strongly labeled with perceptual on and offset times, paying attention to omitting inbetween silences. The amount of isolated sound events, the quality of annotations, and the particular general sound class distinguish NIGENS from other databases.

Wasserstein GAN Can Perform PCA ; Generative Adversarial Networks GANs have become a powerful framework to learn generative models that arise across a wide variety of domains. While there has been a recent surge in the development of numerous GAN architectures with distinct optimization metrics, we are still lacking in our understanding on how far away such GANs are from optimality. In this paper, we make progress on a theoretical understanding of the GANs under a simple lineargenerator Gaussiandata setting where the optimal maximumlikelihood generator is known to perform Principal Component Analysis PCA. We find that the original GAN by Goodfellow et. al. fails to recover the optimal PCA solution. On the other hand, we show that Wasserstein GAN can approach the PCA solution in the limit of sample size, and hence it may serve as a basis for an optimal GAN architecture that yields the optimal generator for a wide range of data settings.

Harmonizing Maximum Likelihood with GANs for Multimodal Conditional Generation ; Recent advances in conditional image generation tasks, such as imagetoimage translation and image inpainting, are largely accounted to the success of conditional GAN models, which are often optimized by the joint use of the GAN loss with the reconstruction loss. However, we reveal that this training recipe shared by almost all existing methods causes one critical side effect lack of diversity in output samples. In order to accomplish both training stability and multimodal output generation, we propose novel training schemes with a new set of losses named moment reconstruction losses that simply replace the reconstruction loss. We show that our approach is applicable to any conditional generation tasks by performing thorough experiments on imagetoimage translation, superresolution and image inpainting using Cityscapes and CelebA dataset. Quantitative evaluations also confirm that our methods achieve a great diversity in outputs while retaining or even improving the visual fidelity of generated samples.

STRIGGER Continual State Representation Learning via SelfTriggered Generative Replay ; We consider the problem of building a state representation model for control, in a continual learning setting. As the environment changes, the aim is to efficiently compress the sensory state's information without losing past knowledge, and then use Reinforcement Learning on the resulting features for efficient policy learning. To this end, we propose STRIGGER, a general method for Continual State Representation Learning applicable to Variational AutoEncoders and its many variants. The method is based on Generative Replay, i.e. the use of generated samples to maintain past knowledge. It comes along with a statistically sound method for environment change detection, which selftriggers the Generative Replay. Our experiments on VAEs show that STRIGGER learns state representations that allows fast and highperforming Reinforcement Learning, while avoiding catastrophic forgetting. The resulting system is capable of autonomously learning new information without using past data and with a bounded system size. Code for our experiments is attached in Appendix.

On the Convergence Rates of Learningbased Signature Generation Schemes to Contain Selfpropagating Malware ; In this paper, we investigate the importance of a defense system's learning rates to fight against the selfpropagating class of malware such as worms and bots. To this end, we introduce a new propagation model based on the interactions between an adversary and its agents who wishes to construct a zombie army of a specific size, and a defender taking advantage of standard security tools and technologies such as honeypots HPs and intrusion detection and prevention systems IDPSes in the network environment. As time goes on, the defender can incrementally learn from the collectedobserved attack samples e.g., malware payloads, and therefore being able to generate attack signatures. The generated signatures then are used for filtering next attack traffic and thus containing the attacker's progress in its malware propagation mission. Using simulation and numerical analysis, we evaluate the efficacy of signature generation algorithms and in general any learningbased scheme in bringing an adversary's maneuvering in the environment to a halt as an adversarial containment strategy.

The Missing Data Encoder CrossChannel Image Completionwith HideAndSeek Adversarial Network ; Image completion is the problem of generating whole images from fragments only. It encompasses inpainting generating a patch given its surrounding, reverse inpaintingextrapolation generating the periphery given the central patch as well as colorization generating one or several channels given other ones. In this paper, we employ a deep network to perform image completion, with adversarial training as well as perceptual and completion losses, and call it the missing data encoder'' MDE. We consider several configurations based on how the seed fragments are chosen. We show that training MDE for random extrapolation and colorization'' MDEREC, i.e. using random channelindependent fragments, allows a better capture of the image semantics and geometry. MDE training makes use of a novel hideandseek'' adversarial loss, where the discriminator seeks the original nonmasked regions, while the generator tries to hide them. We validate our models both qualitatively and quantitatively on several datasets, showing their interest for image completion, unsupervised representation learning as well as face occlusion handling.

Transferable MultiDomain State Generator for TaskOriented Dialogue Systems ; Overdependence on domain ontology and lack of knowledge sharing across domains are two practical and yet less studied problems of dialogue state tracking. Existing approaches generally fall short in tracking unknown slot values during inference and often have difficulties in adapting to new domains. In this paper, we propose a Transferable Dialogue State Generator TRADE that generates dialogue states from utterances using a copy mechanism, facilitating knowledge transfer when predicting domain, slot, value triplets not encountered during training. Our model is composed of an utterance encoder, a slot gate, and a state generator, which are shared across domains. Empirical results demonstrate that TRADE achieves stateoftheart joint goal accuracy of 48.62 for the five domains of MultiWOZ, a humanhuman dialogue dataset. In addition, we show its transferring ability by simulating zeroshot and fewshot dialogue state tracking for unseen domains. TRADE achieves 60.58 joint goal accuracy in one of the zeroshot domains, and is able to adapt to fewshot cases without forgetting already trained domains.

Augmenting Physiological Time Series Data A Case Study for Sleep Apnea Detection ; Supervised machine learning applications in the health domain often face the problem of insufficient training datasets. The quantity of labelled data is small due to privacy concerns and the cost of data acquisition and labelling by a medical expert. Furthermore, it is quite common that collected data are unbalanced and getting enough data to personalize models for individuals is very expensive or even infeasible. This paper addresses these problems by 1 designing a recurrent Generative Adversarial Network to generate realistic synthetic data and to augment the original dataset, 2 enabling the generation of balanced datasets based on heavily unbalanced dataset, and 3 to control the data generation in such a way that the generated data resembles data from specific individuals. We apply these solutions for sleep apnea detection and study in the evaluation the performance of four wellknown techniques, i.e., KNearest Neighbour, Random Forest, MultiLayer Perceptron, and Support Vector Machine. All classifiers exhibit in the experiments a consistent increase in sensitivity and a kappa statistic increase by between 0.007 and 0.182.

Explicitizing an Implicit Bias of the Frequency Principle in Twolayer Neural Networks ; It remains a puzzle that why deep neural networks DNNs, with more parameters than samples, often generalize well. An attempt of understanding this puzzle is to discover implicit biases underlying the training process of DNNs, such as the Frequency Principle FPrinciple, i.e., DNNs often fit target functions from low to high frequencies. Inspired by the FPrinciple, we propose an effective model of linear FPrinciple LFP dynamics which accurately predicts the learning results of twolayer ReLU neural networks NNs of large widths. This LFP dynamics is rationalized by a linearized mean field residual dynamics of NNs. Importantly, the longtime limit solution of this LFP dynamics is equivalent to the solution of a constrained optimization problem explicitly minimizing an FPnorm, in which higher frequencies of feasible solutions are more heavily penalized. Using this optimization formulation, an a priori estimate of the generalization error bound is provided, revealing that a higher FPnorm of the target function increases the generalization error. Overall, by explicitizing the implicit bias of the FPrinciple as an explicit penalty for twolayer NNs, our work makes a step towards a quantitative understanding of the learning and generalization of general DNNs.

Remarks on generic stability in independent theories ; In NIP theories, generically stable Keisler measures can be characterized in several ways. We analyze these various forms of generic stability in arbitrary theories. Among other things, we show that the standard definition of generic stability for types coincides with the notion of a frequency interpretation measure. We also give combinatorial examples of types in NSOP theories that are finitely approximated but not generically stable, as well as phitypes in simple theories that are definable and finitely satisfiable in a small model, but not finitely approximated. Our proofs demonstrate interesting connections to classical results from Ramsey theory for finite graphs and hypergraphs.

The Generic Expansion in Analytic Modified Gravity ; In this Thesis, we treat the problem of the existence of generic perturbations of the regular and singular state in higherorder gravity in cases of vacuum and radiation models that derives from the lagrangian Repsilon R2. We show that there is a regular state of the theory in vacuum in the form of a formal series expansion having the same number of free functions as those required for a general solution of the theory, while this is not true for the case of radiation. This means that there exists an open set in the space of analytic initial data of the theory in vacuum that leads to a regular solution having the correct number of free functions to qualify as a general solution. Further, we show that a singular state of the theory in vacuum cannot be admitted, while in the case of radiation we obtain a particular solution.

Applying Generative Adversarial Networks to Intelligent Subsurface Imaging and Identification ; To augment training data for machine learning models in Ground Penetrating Radar GPR data classification and identification, this thesis focuses on the generation of realistic GPR data using Generative Adversarial Networks. An innovative GAN architecture is proposed for generating GPR Bscans, which is, to the author's knowledge, the first successful application of GAN to GPR Bscans. As one of the major contributions, a novel loss function is formulated by merging frequency domain with time domain features. To test the efficacy of generated Bscans, a real time object classifier is proposed to measure the performance gain derived from augmented BScan images. The numerical experiment illustrated that, based on the augmented training data, the proposed GAN architecture demonstrated a significant increase from 82 to 98 in the accuracy of the object classifier.