In a self-interested setting, agents may strategically hide some connections to make themselves seem to be more important . In this paper, we study the incentive compatible (IC) selectionmechanism to prevent such manipulations . We propose the Geometric Mechanism, which selects an agent with at least 1/2 of the optimal progeny in expectation under the properties of incentive compatibility and fairness .…

## Evaluation of contextual embeddings on less resourced languages

The current dominance of deep neural networks in natural language processing is based on contextual embeddings such as ELMo, BERT, and BERT derivatives . Most existing work focuses on English; in contrast, we present here the firstmultilingual empirical comparison of two ELMo and several monolingual and multilingual BERT models using 14 tasks in nine languages .…

## Hodge theoretic reward allocation for generalized cooperative games on graphs

We define cooperative games on general graphs and generalize Lloyd S.Shapley’s celebrated allocation formula for those games in terms of stochasticpath integral driven by the associated Markov chain on each graph . We then show that the value allocation operator, one for each player defined by thestochastic path integral, coincides with the player’s component game which is the solution to the least squares (or Poisson’s) equation .…

## Spinning Sequence to Sequence Models with Meta Backdoors

We investigate a new threat to neural sequence-to-sequence (seq2seq) models: training-time attacks that cause models to “spin” their output and support acertain sentiment . We introduce the concept of a “meta-backdoor” to explain model-spinningattacks . These attacks produce models whose output is valid and preservescontext, yet also satisfies a meta-task chosen by the adversary (e.g.,…

## Multiple Query Optimization using a Hybrid Approach of Classical and Quantum Computing

Quantum computing promises to solve difficult optimization problems more efficiently than classical computers, but requires fault-tolerant quantum computers with millions of qubits . Hybrid algorithms combining classical and quantum computers are used to overcome errors introduced by today’s quantum computers .…

## Copy and Paste method based on Pose for Re identification

Re-identification (ReID) aims at matching objects in surveillance cameras with different viewpoints . But there is noprocessing method for the ReID task in multiple scenarios at this stage . The CPP is a method based on key point detection, usingcopy and paste, to composite a new semantic image dataset in two different semantic image datasets .…

## Philosophical Specification of Empathetic Ethical Artificial Intelligence

An ethical AI must be capable of inferring unspoken rules, interpreting nuance and context, possess and be able to infer intent, and explain not just its actions but its intent . It can learn what is meant by a sentence and infer the intent of others in terms of its own experiences .…

## Kurtosis of von Neumann entanglement entropy

In this work, we study the statistical behavior of entanglement in quantumbipartite systems under the Hilbert-Schmidt ensemble . We make use of newly observed unsimplifiablesummation bases that lead to a complete cancellation . In addition to providing evidence of the conjectured Gaussian limit of the von Neumann entropy, the obtained formula also provides an improved finite-size approximation to the distribution of the distribution .…

## Domain Generalization under Conditional and Label Shifts via Variational Bayesian Inference

The widely used domain invariant feature learning (IFL) methods relies on aligning the marginal concept shift w.r.t. $p(x|y)$ and the marginal label shift . In this work, we propose a domain generalization (DG) approach to learn on several labeled source domains and transfer knowledge to a target domain that is inaccessible in training .…

## Efficient Neural Causal Discovery without Acyclicity Constraints

Learning the structure of a causal graphical model using both observationaland interventional data is a fundamental problem in many scientific fields . ENCO formulates the graph search as an optimization of independent edge likelihoods, with the edge orientation being modeled as aseparate parameter .…

## Specifying a Game Theoretic Extensive Form as an Abstract 5 ary Relation

This paper specifies an extensive form as a 5-ary relation (i.e. set ofquintuples) which satisfies certain abstract axioms . Each quintuple is understood to list a player, a situation (e.g. information set), a decisionnode, a decision node, an action, and a successor node .…

## Dialogue Object Search

We envision robots that can collaborate and communicate seamlessly with humans . We introduce a new task,dialogue object search: A robot is tasked to search for a target object in a human environment while engaging in a “video call” with a remote human who has additional but inexact knowledge about the target’s location .…

## 3D Shape Generation with Grid based Implicit Functions

Previous approaches to generate shapes in a 3D setting train a GAN on thelatent space of an autoencoder (AE) This produces convincing results, but it has two major shortcomings . To remedy these issues, we propose to train the GAN .…

## To Ship or Not to Ship An Extensive Evaluation of Automatic Metrics for Machine Translation

Automatic metrics are commonly used as the exclusive tool for declaring thesuperiority of one machine translation system’s quality over another . We investigate which metrics have the highest accuracy to makesystem-level quality rankings for pairs of systems . We show that the sole use of BLEUnegatively affected the past development of improved models.…

## Bandit Quickest Changepoint Detection

Detecting abrupt changes in temporal behavior patterns is of interest in many industrial and security applications . Abrupt changes are often local andobservable primarily through a well-aligned sensing action (e.g., a camera with a narrow field-of-view) Due to resource constraints, continuous monitoring of all of the sensors is impractical .…

## Target Oriented Fine tuning for Zero Resource Named Entity Recognition

Zero-resource named entity recognition (NER) severely suffers from datascarcity in a specific domain or language . Most studies on zero-resource NERtransfer knowledge from various data by fine-tuning on different auxiliary tasks . In this paper, we tackle the problem by transferring knowledge from three aspects, i.e.,…

## Impacts Towards a comprehensive assessment of the book impact by integrating multiple evaluation sources

The surge in the number of books published makes the manual evaluation methods difficult to efficiently evaluate books . The use of books’ citationsand alternative evaluation metrics can assist manual evaluation and reduce the cost of evaluation . However, relying on a single resource for book assessment may lead to the risk that theevaluation results cannot be obtained due to the lack of the evaluation data, especially for newly published books .…

## A Logic of Expertise

In this paper we introduce a simple modal logic framework to reason about theexpertise of an information source . In the framework, a source is an expert on a proposition $p$ if they are able to correctly determine the truth value of$p$ in any possible world .…

## Back Translated Task Adaptive Pretraining Improving Accuracy and Robustness on Text Classification

Language models (LMs) pretrained on a large text corpus and fine-tuned on a downstream task becomes a de factotraining strategy for several natural language processing (NLP) tasks . We propose aback-translated task-adaptive pretraining (BT-TAPT) method that increases the amount of task-specific data for LM re-pretraining by augmenting the task data .…

## The Public Good index for games with several levels of approval in the input and output

The Public Good index is a power index for simple games introduced by Holler and Packel . Some authors also speak of the Holler–Packel index . Here we generalize the ideas to games with several levels of approval in the input and output .…

## Theoretical foundations and limits of word embeddings what types of meaning can they capture

Measuring meaning is a central problem in cultural sociology and word embeddings may offer powerful new tools to do so . But like any tool, they buildon and exert theoretical assumptions . In certain ways, word embedding methods are vulnerable to the same, enduring critiques of these premises .…

## Did the Cat Drink the Coffee Challenging Transformers with Generalized Event Knowledge

Computational approaches have access to the information about thetypicality of entire events and situations described in language . The evaluation of these models was performed incomparison with SDM, a framework specifically designed to integrate events insentence meaning representations . Our results show that TLMs can reach performances that are comparable to those achieved by SDM .…

## Neural Ordinary Differential Equation Model for Evolutionary Subspace Clustering and Its Applications

In multi-dimensional time series analysis, a task is to conductevolutionary subspace clustering, aiming at clustering temporal data according to their evolving low-dimensional subspace structures . We demonstrate that this method can not onlyinterpolate data at any time step, but also achieve higher accuracy than other state-of-the-art evolutionarysubspace-cluing methods .…

## Griddings of permutations and hardness of pattern matching

We study the complexity of the decision problem known as Permutation PatternMatching . The input of PPM consists of a pair of permutations, and the goal is to decide whether $tau$ contains a subpermutation . On general inputs, PPM is known to be NP-complete by a result of Bose, Buss and Lubiw .…

## FNetAR Mixing Tokens with Autoregressive Fourier Transforms

In this note we examine the autoregressive generalization of the FNetalgorithm . Self-attention layers from the standard Transformerarchitecture are substituted with a trivial sparse-uniformsampling procedure based on Fourier transforms . Using the Wikitext-103 benchmark, FNetAR retains state-of-the-art performance (25.8 ppl) on thetask of causal language modeling .…

## Fourier growth of structured mathbb F _2 polynomials and applications

We analyze the Fourier growth of various well-studied classes of “structured”$\mathbb{F}_2$-polynomials . This study is motivated by applications inpseudorandomness, in particular recent results and conjectures due to[CHHL19,CHLT19,CGLSS20] We show that any symmetric degree-$d$ $p$ has $L_1$ Fourier weight at level $k$ and this is tight for any constant $k$.…

## MFGNet Dynamic Modality Aware Filter Generation for RGB T Tracking

MFGNet aims to boost the message communication between visible and thermaldata by adaptively adjusting the convolutional kernels for various input images . To address issues caused by heavy occlusion, fast motion, and out-of-view, we propose to conduct a joint local and global search byexploiting a new direction-aware target-driven attention mechanism .…

## A reinforcement learning approach to resource allocation in genomic selection

Genomic selection (GS) is a technique that plant breeders use to select individuals to mate and produce new generations of species . Allocation of resources is a key factor in GS . Inspired by recent advances in reinforcement learning forAI problems, we develop a reinforcement learning-based algorithm .…

## Shedding some light on Light Up with Artificial Intelligence

The Light-Up puzzle, also known as the AKARI puzzle, has never been solved using modern artificial intelligence (AI) methods . This project is an effort to apply new AI techniques for solving the Light-up puzzle faster and more computationallyefficient . The algorithms explored for producing optimal solutions include hillclimbing, simulated annealing, feed-forward neural network (FNN), and CNN, and an evolutionary theory algorithm .…

## Distributed Saddle Point Problems Under Similarity

We study solution methods for (strongly-)convex-(strongly)-concaveSaddle-Point Problems (SPPs) over networks of two type – master/workers (thuscentralized) and meshed (thus decentralized) networks . We establish lower complexity bounds for a fairlygeneral class of algorithms solving the SPP . We then propose algorithms matching the lower bounds over either types of networks (up tolog-factors) We assess the effectiveness of the proposed algorithms on arobust logistic regression problem .…

## HARP Net Hyper Autoencoded Reconstruction Propagation for Scalable Neural Audio Coding

An autoencoder-based codec employs quantization to turn its bottleneck layeractivation into bitstrings . To circumvent this issue, we employ additional skipconnections between the corresponding pair of encoder-decoder layers . We empirically verify that the proposedhyper-autoencoded architecture improves perceptual audio quality compared to an ordinary autoencoder baseline .…

## Fourier Reflexive Partitions Induced by Poset Metric

Let $H$ be the cartesian product of a family of finite abeliangroups indexed by a finite set . A given poset (i.e., partially orderedset) gives rise to a posetmetric on a given set . We prove that if$P} is Fourier-reflexive, then its dualpartition $Lambda$ coincides with the partition of the dual poset of $\mathbf{H$ .…

## Typing assumptions improve identification in causal discovery

Under assumptions about the data-generative process, the causal graph can often be identified up to anequivalence class . Proposing new realistic assumptions to circumscribe suchequivalence classes is an active field of research . In this work, we propose anew set of assumptions that constrain possible causal relationships based on the nature of the variables .…

## High dimensional expansion implies amplified local testability

In this work we show that high dimensional expansion implies locally testablecode . We define a notion that we callhigh-dimensional-expanding-system (HDE-system) We show that a code that can be modelled over HDE-system islocally testable . This implies that high-dimensional expansion phenomenon solely implies local testability of codes .…

## The Optimality of Upgrade Pricing

We consider a multiproduct monopoly pricing model . We provide sufficient conditions under which the optimal mechanism can be implemented via upgradepricing . The first set ofconditions is given by a weak version of monotonicity of types and virtualvalues . The second set of conditions establishes the optimality of upgrade pricing for type spaces withmonotone marginal rates of substitution .…

## Lower Bounds for Maximally Recoverable Tensor Code and Higher Order MDS Codes

Maximally Recoverable (MR) TensorCodes, introduced by Gopalan et al., are tensor codes which can correct everyerasure pattern that is theoretically possible to correct . Tensor codes are useful in distributed storage because a single erasure can be correctedquickly either by reading its row or column .…

## Distributed Asynchronous Policy Iteration for Sequential Zero Sum Games and Minimax Control

We introduce a contractive abstract dynamic programming framework and relatedpolicy iteration algorithms . These algorithms are specifically designed for sequential zero-sumgames and minimax problems with a general structure . The advantage of our algorithms over alternatives is that they resolve some long-standing convergence difficulties of the “natural” policyiteration algorithm, which have been known since the Pollatschek and Avi-Itzhakmethod [PoA69] for finite-state Markov games .…

## Evaluation of In Person Counseling Strategies To Develop Physical Activity Chatbot for Women

Artificial intelligence chatbots are the vanguard in technology-based intervention to change people’s behavior . To develop intervention chatbots, the first step is to understand natural language conversation strategies in humanconversation . This work lays the foundation for developing a personalized physical activity intervention bot .…

## Abstract Reasoning via Logic guided Generation

Abstract reasoning, i.e., inferring complicated patterns from givenobservations, is a central building block of artificial general intelligence . We propose logic-guided generation (LoGe), a novelgenerative DNN framework that reduces abstract reasoning as an optimization problem in propositional logic . LoGe is composed of three steps: extractpropositional variables from images, reason the answer variables with a logiclayer, and reconstruct the answer image from the variables .…

## Randomized Online Algorithms for Adwords

The general adwords problem has remained largely unresolved . We define asubcase called $k-TYPICAL, $k \in \Zplus$ as follows: the total budgetof all the bidders is sufficient to buy $k$ bids for each bidder . We also giverandomized online algorithms for other special cases of adwords .…

## On the Certified Robustness for Ensemble Models and Beyond

Deep neural networks (DNN) are vulnerable to adversarial examples, which aim to mislead DNNs by adding perturbations with small magnitude . In terms of thecertified robustness the standard ensemble models only achieve marginal improvement compared to a single model . Inspired by the theoretical findings, we propose the lightweight Diversity Regularized Training (DRT) to train certifiably robustensemble ML models .…

## Joint Optimization of Preamble Selection and Access Barring for Random Access in MTC with General Device Activities

Most existing random access schemes for MTC simply adopt a uniform preambleselection distribution, irrespective of the underlying device activitydistributions . We investigate three cases of the general joint device activity distribution . We formulate theaverage, worst-case average, and sample average throughput maximization maximization problems .…

## Domain Generalization under Conditional and Label Shifts via Variational Bayesian Inference

The widely used domain invariant feature learning (IFL) methods relies on aligning the marginal concept shift w.r.t. $p(x|y)$ and the marginal label shift . In this work, we propose a domain generalization (DG) approach to learn on several labeled source domains and transfer knowledge to a target domain that is inaccessible in training .…

## Improving Polyphonic Sound Event Detection on Multichannel Recordings with the Sørensen Dice Coefficient Loss and Transfer Learning

The S.rensen–Dice Coefficient has recently seen rising popularity as aloss function (also known as Dice loss) due to its robustness in tasks where the number of negative samples significantly exceeds that of positive samples . Conventional training of polyphonic sound event detection systemswith binary cross-entropy loss often results in suboptimal detectionperformance as the training is often overwhelmed by updates from negativesamples .…

## Equidistant Linear Codes in Projective Spaces

Linear codes in projective space $mathbb{P}_q(n)$ were first considered by Braun,Etzion and Vardy . We establish that the normalizedminimum distance of a linear code is maximum if and only if it is equidistant . We prove that the upper bound on the size of such class of linear codes is $2^n$ when $q=2$ as conjectured by Braun et al.…

## A Framework for Imbalanced Time series Forecasting

Time-series forecasting plays an important role in many domains, such as wind power, stock market fluctuations, or motor overheating . In some of these tasks, some particular moments often are underrepresented in thedataset, resulting in a problem known as imbalanced regression .…

## Evaluation of In Person Counseling Strategies To Develop Physical Activity Chatbot for Women

Artificial intelligence chatbots are the vanguard in technology-based intervention to change people’s behavior . To develop intervention chatbots, the first step is to understand natural language conversation strategies in humanconversation . This work lays the foundation for developing a personalized physical activity intervention bot .…

## Fabrication Aware Reverse Engineering for Carpentry

We propose a novel method to generate fabrication blueprints from images of carpentered items . We demonstrate our method on a variety of wooden objects and furniture . We can automatically obtain designs that are both easy to edit and accurate recreations of the ground truth .…

## Confidence Aware Scheduled Sampling for Neural Machine Translation

Scheduled sampling is an effective method to alleviate the exposure biasproblem of neural machine translation . It simulates the inference scene by replacing ground-truth target input tokens with predicted ones during training . Despite its success, its critical schedule strategies are merelybased on training steps, ignoring the real-time model competence .…

## Graph Based Learning for Stock Movement Prediction with Textual and Relational Data

Multi-Graph Recurrent Network for Stock Forecasting (MGRN) combines textual and relational information extracted from financial news and other financial data . One news regarding one stock can quickly impact the prices of other stocks . This architecture allows to combine the textualsentiment from news and multiple relational data extracted from other data .…