Transfer Learning for Human & AI

Non-isotropic geometries are of interest to low-dimensional topologists,physicists and cosmologists . However, they are challenging to comprehend andvisualize . We present novel methods of computing real-time native geodesicrendering . Our methods can be applied not only tovisualization, but also are essential for potential applications in machinelearning and video games .…

We show that solving games with imperfect recall is as hard as solving certain problems of the firstorder theory of reals . We establish square root sum hardness even for the class of A-loss games . On the positive side, we find restrictions on games and strategies motivated by Bridge bidding that give polynomial-timecomplexity .…

N cooperative but non-communicating players where each plays one outof M arms for T turns . Players have different utilities for each arm,representable as an NxM matrix . These utilities are unknown to the players . In each turn players select an arm and receive a noisy observation of their utility for it .…

We provide prophet inequality algorithms for online weighted matching ingeneral (non-bipartite) graphs . The weight of each edge is drawn independently from an a-priori known probability distribution . We introduce a novel unified framework of batched prophet inequalities that captures online settings where elements arrive in batches .…

In this paper, we consider multi-agent learning via online gradient descent in a class of games called $lambda$-cocoercive games . We characterize the finite-time last-iterate convergence rate for joint OGD learning on these games . Further, we develop a fully adaptive algorithm that does not require any knowledge of problem parameter (e.g.…

The paper develops a logical understanding of processes for signature of legal contracts . It is motivated by applications to legal recognition of smartcontracts on blockchain platforms . A number of axioms and rules of inference are developed that can be used to justify a “meeting of the minds” precondition for contract formation from the fact that certain content has been signed .…

We consider a decentralized stochastic learning problem where data points are distributed among computing nodes communicating over a directed graph . As themodel size gets large, decentralized learning faces a major bottleneck that is the heavy communication load due to each node transmitting large messages to its neighbors .…

Large crowds are common for large crowds to gather to attend games, exhibitions, rallies, and other events . However, congestion created by large crowds has resulted in hazardous incidents across the world . Developments in information technology can provide new means to disseminate public information, thuschanging human behavior in situations of danger and duress .…

We consider a system to optimize duration of traffic signals using multi-agent deep reinforcement learning and Vehicle-to-Everything (V2X) communication . A learning agent trafficlight gets information along its lanes within a circular V2X coverage . Weevaluate multi-agents through the Simulation of Urban MObility (SUMO)simulator .…

In addition to seamless connectivity and smartness, the objects in the Internet of Things (IoT) are expected to have the social capabilities . In this paper, an intuitive paradigmof social interactions between these objects are argued and modeled . The impact of social behavior on the interaction pattern of social objects is studied taking Peer-to-Peer (P2P) resource sharing as an example application .…

We present a full-program induction technique for proving (a sub-class of)quantified as well as quantifier-free properties of programs manipulatingarrays of parametric size N . Instead of inducting over individual loops, ourtechnique inducts over the entire program (possibly containing multiple loops) Significantly, this does not requiregeneration or use of loop-specific invariants .…

The aim of this work is to provide a family of theories for spatial change in general and for motion of spatial scenes in particular . We consider a spatio-temporalisation MTALC(Dx), of the well-knownALC family of Description Logics . The roles split into m+n immediate-successor(accessibility) relations, which are serial, irreflexive and antisymmetric .…

The aim of this work is to thoroughly investigate Buchi automata augmentedwith spatial constraints . The input trees of such an automaton are infinitek-ary Sigma-trees, with the nodes standing for time points . Theconstraints, from an RCC8-like spatial Relation Algebra (RA) x, are used to impose spatial constraints on objects of the spatial scene .…

Promising federated learning coupled with Mobile Edge Computing (MEC) is considered as one of the most promising solutions to the AI-driven serviceprovision . In MEC, edgenodes would not like to voluntarily participate in learning, and they differ inthe provision of multi-dimensional resources .…

A well-known axiom for proportional representation is Proportionality ofSolid Coalitions . We characterize committees satisfying PSC as possible outcomes of the Minimal Demand rule .…

The concept of a uniform interpolant for a quantifier-free formula from agiven formula with a list of symbols, while well-known in the logic literature,has been unknown to the formal methods and automated reasoning community . The first algorithm is non-deterministic andgenerates a uniforminterpolant expressed as a disjunction of conjunction ofliterals .…

Concurrent Kleene Algebra (CKA) extends basic Kleene algebra with a parallelcomposition operator . However, CKA fundamentally misses tests, which are needed to model standardprogramming constructs such as conditionals and $\mathsf{while}$-loops . In this paper we provide a solution in the form of ConcurrentKleeneAlgebra with Observations .…

Automatic Chord Estimation (ACE) is a fundamental task in Music InformationRetrieval (MIR) The task consists of segmenting a music recording or score and assigning a chord label to each segment . DECIBEL improves all tested state-of-the-art ACEmethods by over 3 percent on average .…

The framework takes multiple representations as the input to train themodels in parallel . The complementary information provided by different representations is shared by knowledge distillation . The proposed approach can improve the classificationperformance and achieve state-of-the-art results on both acoustic sceneclassification tasks and general audio tagging tasks .…

Devirtualization is a compiler optimization that replaces indirect (virtual)function calls with direct calls . It is particularly effective in languages such as Java or C++, in which virtual methods are abundant . We present a novel abstract model to express the lifetimes of C++ dynamicobjects and invariance of virtual table pointers in the LLVM intermediaterepresentation .…

We show that, while Minesweeper is NP-complete, its hyperbolic variant is inP . Our proof does not rely on the rules of the game, but is valid for any puzzle based on satisfying local constraints .…

The geometric rank is an upper bound on the subrank of tensors and the independence number of hypergraphs . We prove it is smaller than the slice rank of Tao, and relategeometric rank to the analytic rank of Gowers and Wolf in an asymptotic fashion .…

Finite-domain constraint satisfaction problems are either solvable by Datalog, or not even expressible in fixed-point logic with counting . For infinite-domain CSPs, the situation is more complicated even if the template structure of the CSP is model-theoretically tame . We prove that there is no Maltsev condition that characterizes Dalog already for theCSPs of first-order reducts of (Q; <) We also prove that many of the equivalent conditions in the finite fail to capture expressibility in Datalogs or fixed point logic . The border between the two regimes coincides with an important dichotomy in universalalgebra; in particular, the border can be described by a strong height-one …

We show that any automatic sequence can be separated into a structured part and a Gowers uniform part in a way that is considerably more efficient than the Arithmetic Regularity Lemma . For sequences produced bystrongly connected and prolongable automata, the structured part is rationallyalmost periodic, while for general sequences the description is marginally more complicated .…

We show that if the added language $L$ has a neutral letter, the resulting logic is necessarily undecidable . A corollary is that the $\omega$-regular languagesare the only decidable Boolean-closed full trio over $lions .…

Images can be represented by mathematical function using wavelets . Wavelet can be manipulated (shrink/expand) by applying some values to its function . Application of wavelet in images processing haslarger scope as proved . There arevarious wavelet image compression techniques. This research paper focused oncomparison of only two techniques i.e.…

In this paper, we introduce a two-player impartial game on graphs, called a feedback game, which is a variant of the generalized geography . We study the feedback game on Eulerian graphs . In particular, we show that thePSPACE-completeness of the game determines the winner .…

We study an optimization problem capturing a core operational question for online retailing platforms . We model theprobability that a user clicks on an element from a subset of products as amonotone submodular function of this set . We also assume that the patiencelevel of the users, or the number of items they are willing to observe beforeclicking on one or abandoning the search, is a given random variable .…

When voter preferences are known in an incomplete (partial) manner, winnerdetermination is commonly treated as identification of necessary and possible winners . Yet, the outcome of an election might go beyond the absolute winners to the top-$k$ winners, as in the case of committeeselection, primaries of political parties, and ranking in recruiting .…

Study: Community members recruited by law enforcement as informants can assist patrols . We introduce novel two-stage security game model for community engagement . We also provide a novel algorithm to find the optimal defender strategy against level-$\kappa$ ($\infty$) attackers .…

Generative adversarial networks (GANs) represent a zero-sum game between twomachine players, a generator and a discriminator, designed to learn the distribution of data . While GANs have achieved state-of-the-art performance in benchmark learning tasks, GAN minimax optimization still poses greattheoretical and empirical challenges .…

We investigate common knowledge equilibrium of separable (or parity) andtotally symmetric Boolean securities in distributed information market . Wetheoretically show that clearing price converges to the true value when acommon prior probability distribution of information of each player satisfiessome conditions .…

Enterprises are increasingly concerned about adversaries that slowly and deliberate exploit resources over the course of months or even years . New networking technology increases the possibility of passive network reconnaissance, which can be undetectable by defenders . In this paper, we propose Snaz, atechnique that uses deceptively crafted honey traffic to confound the knowledge gained through passive reconnaissance .…

The PBS-calculus is inspired by quantum optics, in particular the polarising beam splitter (PBS for short) We formalise the syntax and semantics of the PBS-diagrams . We equip the language with an equationaltheory, which is proved to be sound and complete .…

In this work we tackle a kinetic-like model of opinions dynamics in anetworked population endued with a quenched plurality and polarization . We consider pairwise interactions that are restrictive, which ismodeled with a smooth bounded confidence . Our results show the interestingemergence of nonequilibrium hysteresis and heterogeneity-assisted ordering .…

The decreased size and cost of Unmanned Aerial Vehicles (UAVs) has enabled the use of swarms of unmanned autonomous vehicles to accomplish a variety of tasks . This work investigates the influencesthat can compromise the functionality of an autonomous swarm leading to cascading vulnerabilities .…

Existing sketch-analysis work studies sketches depicting static objects or scenes . In this work, we propose a novel cross-modal retrieval problem offine-grained instance-level sketch-based video retrieval (FG-SBVR) We show that this model significantly outperforms anumber of existing state-of-the-art models designed for video analysis.…

Quality assessment of omnidirectional images has become increasingly urgent due to the rapid growth of virtual reality applications . We propose a novel Viewportoriented Graph Convolution Network (VGCN) for blind omniddirectional imagequality assessment (IQA) The proposed graph is inspired by the characteristics of the human vision system (HVS) and theviewing process of Omnidirectionable contents .…

Automatic amortized resource analysis (AARA) is a type-based technique forinferring concrete (non-asymptotic) bounds on a program’s resource usage . The soundness of exponential AARA is proved with respect to anoperational cost semantics . A key idea is theuse of the Stirling numbers of the second kind as the basis of potentialfunctions, which play the same role as the binomial coefficients in polynomialAARA .…

We present a framework for symbolically executing and model checkinghigher-order programs with external (open) methods . We combine traditional symbolic executiontechniques with operational game semantics to build a symbolic executionsemantics that captures arbitrary external behaviour . This yields a bounded technique by imposing bounds on the depth of recursion and callbacks .…

We propose the total staggered quantum walk model and the total tessellationcover of a graph . We establish bounds on $T_t(G)$ which is thesmallest number of tesselations required . The $k$-total tessesellability problem aims to decide whether agiven graph $G$ has $K$ (K) = k) We establish hardness results for bipartite graphs, line graphs of triangle-free graphs, universal graphs, planar graphs, and$(2,1)$-chordal graphs .…

We study the problem of finding $K$ collision pairs in a random function $f :[N] \rightarrow [N]$ by using a quantum computer . We prove that the number ofqueries to the function in the quantum random oracle model must increase significantly when the size of the available memory is limited .…

We define and study a class of subshifts of finite type (SFTs) defined by afamily of allowed patterns of the same shape . For such an SFT, a locally legal pattern of convex shape is globally legal, and there is a measure that samples uniformly on all convexsets .…

In this paper two algorithms solving circuit satisfiability problem oversupernilpotent algebras are presented . The first one is deterministic and is faster than the fastest previous algorithm presented by Aichinger . The second one is probabilistic with linear time complexity .…

In arXiv:1710.08163 a generalization of Boolean circuits to arbitrary finitealgebras had been introduced and applied to sketch P versus NP-completeborderline for circuits satisfiability . However the problem for nilpotent (which had not been shown to be NP-hard) but not supernilpotent .…

The study of the orthogonality dimension of graphs ismotivated by various application in information theory and in theoreticalcomputer science . We prove that there exists a constant $c$ such that for everysufficiently large integer $t$, it is $t$-hard to decide whether theorthog on an input graph is at most $t/2-c$ or at least $3/2 .…

The class $FORMULA[s] \circ \mathcal{G}$ consists of Boolean functionscomputable by size-$s$ de Morgan formulas . We give lower bounds and (SAT, Learning,and PRG) algorithms . We show: (1) The Generalized Inner Product function $GIP^k_n$ cannot be computed in $formulA[n^{1.99}]\circ . (2) There is a PRG of seed length $n/2 + O\left(\sqrt{s} \cdotR^{(2)(\mathcal {G}))\right) The Minimum Circuit Size Problem is not in $FORMula[n\circ XOR$.…

Regular expressions with capture variables, also known as “regex formulas,” extract relations of spans (intervals identified by their start and endindices) from text . Fagin et al. introduced regular documentspanners which are the closure of regex formulas under Relational Algebra .…

In interface theory and model-based testing withinputs and outputs, conjunctive operators have been introduced . In the theory of alternatingautomata, conjunction and non-determinism are core aspects . Alternating automata have not been considered in the context of inputs and outputs .…

We present the notion of stateful priorities for imposing preciserestrictions on system actions, in order to meet safety constraints . Given a system modeled as a network of discrete automata and anerror constraint, we present algorithms which use those inputs to synthesizestateful priorities .…