A tangle is a connected topological space constructed by gluing several copies of the unit interval $[0, 1] We explore which tangles guaranteeenvy-free allocations of connected shares for n agents . We show that exactly six tangles are stringable; these guarantee EFconnected allocations for any number of agents, and their associatedtopological classes contain only Hamiltonian graphs .…

## cuFINUFFT a load balanced GPU library for general purpose nonuniform FFTs

GPU-based CUDA library for type 1 (nonuniform touniform) and type 2 (uniform to nonuniform) transforms in dimensions 2 and 3,in single or double precision . It achieves high performance for a given accuracy, regardless of the distribution of non-uniform points, via cache-aware point reordering and load-balanced blocked spreading in sharedmemory .…

## Quantifying environment and population diversity in multi agent reinforcement learning

Generalization is a major challenge for multi-agent reinforcement learning . Training with a diverse set of co-players strengthens agent performance in some (but not all) cases . Results demonstrate that population size and intrinsic motivation are both effective methods of .generating…

## Large deviations for the largest eigenvalue of Gaussian networks with constant average degree

Large deviation behavior of the largest eigenvalue of Gaussiannetworks (Erd\H{o}s-R\’enyi random graphs with i.i.d.Gaussian weights on the edges) has been the topic of considerable interest . We show that conditioned on the uppertail event, with high probability, a unique maximal clique emerges with a veryprecise $delta$ dependent size .…

## A Lazy Approach for Efficient Index Learning

Learned indices using neural networks outperform traditional indices such as B-trees in both query time and memory . Learning the distribution of a large dataset can be expensive, and updatinglearned indices is difficult . We propose a novel approach toreuse these pre-trained models for a new (real) dataset .…

## Boosting Low Resource Biomedical QA via Entity Aware Masking Strategies

Biomedical question-answering (QA) has gained increased attention for its ability to provide users with high-quality information from a vastscientific literature . Finetuning large models can be costly and timeconsuming, often yielding limited benefits when adapting to specific themes of specific domains, such as the COVID-19 literature .…

## Significant Improvements over the State of the Art A Case Study of the MS MARCO Document Ranking Leaderboard

Leaderboards are a ubiquitous part of modern research in applied machinelearning . By design, they sort entries into some linear order, where the top-scoring entry is recognized as the “state of the art” (SOTA) Such pronouncements,however, are almost never qualified with significance testing .…

## User Inspired Posterior Network for Recommendation Reason Generation

Recommendation reason generation plays a vital role in attracting customers’ attention and improving user experience . A simple and effective way is to extractkeywords directly from the knowledge-base of products, i.e., attributes ortitle, as the recommendation reason . But generating recommendation reason from product knowledge doesn’t naturally respond to users’ interests .…

## Scheduling with Contact Restrictions A Problem Arising in Pandemics

We study a scheduling problem arising in pandemic times where jobs should keep sufficient distance during transit times to machines . The goal is to find a conflict-free schedule of minimum makespan . We show that, unless P=NP, the problem does not allow for a constant factorapproximation even for identical jobs .…

## Nominal Unification and Matching of Higher Order Expressions with Recursive Let

A sound and complete algorithm for nominal unification of higher-orderexpressions with a recursive let is described . We also explore specializations like nominalletrec-matching for expressions, for DAGs, and for garbage-free expressions . We show that it also runs in nondeterministic polynomial time .…

## Moral Decision Making in Medical Hybrid Intelligent Systems A Team Design Patterns Approach to the Bias Mitigation and Data Sharing Design Problems

Increasing automation in the healthcare sector calls for a HybridIntelligence (HI) approach to closely study and design the collaboration of humans and machines . Ensuring that medical HI systems’decision-making is ethical is key . The use of Team Design Patterns (TDPs) canadvance this goal by describing successful and reusable configurations ofdesign problems in which decisions have a moral component, as well as facilitating communication in multidisciplinary teams designing HI systems .…

## Fast deterministic algorithms for computing all eccentricities in hyperbolic Helly graphs

A graph is Helly if every family of pairwise intersecting balls has anonempty common intersection . The class of Helly graphs is the discreteanalogue of the class of hyperconvex metric spaces . It is also known that everygraph isometrically embeds into a Helly graph, making the latter an importantclass of graphs in Metric Graph Theory .…

## Decidability for Sturmian words

We show that the theory of Sturmian words over Presburger arithmetic is decidable . We use a general adder recognizing addition in Ostrowski numeration systems by Baranwal, Schaeffer and Shallit . We then deduce the decidability of the first-order theory of the class of such structures .…

## ReGraphX NoC enabled 3D Heterogeneous ReRAM Architecture for Training Graph Neural Networks

Graph Neural Network (GNN) is a variant of Deep Neural Networks (DNNs)operating on graphs . GNNs are more complex compared to traditional DNNs as they exhibit features of both DNN and graphapplications . ReGraphX outperforms conventional GPUs by up to 3.5X (on anaverage 3X) in terms of execution time, while reducing energy consumption by as much as 11X .…

## Follow the Regularized Leader Routes to Chaos in Routing Games

We study the emergence of chaotic behavior of Follow-the-Regularized Leader(FoReL) dynamics in games . We focus on the effects of increasing the populationsize or the scale of costs in congestion games . Although FoReL dynamics can be strange andnon-equilibrating, we prove that the time average still converges to an exact equilibrium for any choice of learning rate and any scale of cost .…

## Finding the Gap Neuromorphic Motion Vision in Cluttered Environments

In the fly brain, motion-sensitive neurons indicate the presence of nearby objects and directional cues are integrated within an areaknown as the central complex . The agent’s manoeuvres result from a closed action-perception loop implementingprobabilistic decision-making processes . This loop-closure is thought to havedriven the development of neural circuitry in biological agents since the Cambrian explosion .…

## Metatheory jl Fast and Elegant Algebraic Computation in Julia with Extensible Equality Saturation

Metatheory.jl is a lightweight and performant general purposesymbolics and metaprogramming framework . It is meant to simplify the act of writing complex Julia metaprograms and to significantly enhance Julia with a nativeterm rewriting system, based on state-of-the-art equality saturationtechniques, and a dynamic first class Abstract Syntax Tree (AST) patternmatching system that is dynamically composable in an algebraic fashion, taking advantage of the language’s powerful reflection capabilities .…

## Data provenance curation and quality in metrology

Data metrology has emerged as an important requirement for UK National Physical Laboratory and other national metrologyinstitutes . Data provenance and data curation are key components for emergingunderstanding of data metrology . To date provenance research has had limited visibility to or uptake in metrology, authors say .…

## C11Tester A Race Detector for C C Atomics Technical Report

C11Tester is a race detector for the C/C++ memory model . It can explore executions in a larger fragment of the memory model than previous race detector tools . It uses a new constraint-based algorithm to implementmodification order that is optimized to allow C11 tester to make decisions interms of application-visible behaviors .…

## k critical trees and k minimal trees

A prime graph $G$ is $k$-minimal if there is no proper induced subgraph of$G$ containing $X$ is prime . A graph is critical if $G – x$ is critical; $G-x$ is also critical . This research paper attempts to answer I.Boudabbous’s…

## A Federated Data Driven Evolutionary Algorithm

Data-driven evolutionary optimization has witnessed great success in solving complex real-world optimization problems . However, existing data-drivenoptimization algorithms require that all data are centrally stored, which is not always practical and may be vulnerable to privacy leakage and security threats if the data must be collected from different devices .…

## A Qualitative Theory of Cognitive Attitudes and their Change

We present a general logical framework for reasoning about agents’ cognitiveattitudes of both epistemic type and motivational type . We show that it allows us to express a variety of relevant concepts for qualitative decision theory . We also present twoextensions of the logic, one by the notion of choice and the other by dynamicoperators for belief change and desire change .…

## TableLab An Interactive Table Extraction System with Adaptive Deep Learning

TableLab lets users and models seamlessly work together to quicklycustomize high-quality extraction models . TableLab detects tables with similar structures(templates) by clustering embeddings from the extraction model . It then selects a few representative table examples already extracted with a pre-trained base deep learning model .…

## A Cooperative Memory Network for Personalized Task oriented Dialogue Systems with Incomplete User Profiles

There is increasing interest in developing personalized Task-oriented Dialogue Systems . Previous work on personalized TDSs often assumes thatcomplete user profiles are available for most or even all users . We propose a Cooperative Memory Network (CoMemNN) that has a novel mechanism to graduallyenrich user profiles as dialogues progress .…

## Federated Multi armed Bandits with Personalization

A general framework of personalized federated multi-armed bandits (PF-MAB) is proposed . A mixed bandit learning problem flexibly balances generalization and personalization . Theoretical analysis proves that PF-UCB achieves an$O(\log(T)$ regret regardless of the degree of personalization, and has asimilar instance dependency as the lower bound .…

## Efficient Competitions and Online Learning with Strategic Forecasters

Winner-take-all competitions in forecasting and machine-learning suffer fromdistorted incentives . Witkowskiet al. identified this problem and proposed ELF,a truthful mechanism to select a winner . We show that from a pool of $n$forecasters, ELF requires $\Theta(n\log n)$ events or test data points toselect a near-optimal forecaster with high probability .…

## DFAC Framework Factorizing the Value Function via Quantile Mixture for Multi Agent Distributional Q Learning

In fully cooperative multi-agent reinforcement learning (MARL) settings, the environments are highly stochastic due to the partial observability of each agent and the continuously changing policies of the other agents . We propose a Distributional Value FunctionFactorization (DFAC) framework to generalize expected value functionfactorization methods to their DFAC variants .…

## Unambiguous DNFs from Hex

We exhibit an unambiguous k-DNF formula that requires CNF widthOmega~(k^{1.5) Our construction is inspired by the board game Hex and it isvastly simpler than previous ones . Our result is known to imply several other improved separations in query andcommunication complexity .…

## Metropolis Walks on Dynamic Graphs

We study a lazy Metropolis walk of Nonaka, Ono, Sadakane, and Yamashita(2010) which is a weighted random walk using local degree information . We show that this walk is robust to an edge-changing in dynamic networks . We give upper bounds of those times for any reversible random walks with a time-homogeneous stationary distribution.…

## Quasi Monte Carlo Software

This article describes some of the better quasi-Monte Carlo (QMC) software available . We highlight the key software components required to approximate multivariate integrals . We have combined these components in QMCPy, a Python open source library . We hope to draw the support of the QMC community .…

## A Bayesian Approach for Inferring Sea Ice Loads

The Earth’s climate is rapidly changing and some of the most drastic changes can be seen in the Arctic . As the Arctic climate continues to change, gathering in situ seaice measurements is increasingly important for understanding the complex evolution of the Arctic ice pack .…

## THIA Accelerating Video Analytics using Early Inference and Fine Grained Query Planning

Researchers have proposed twotechniques for lowering the computational overhead associated with the underlying deep learning models . THIA, a video analytics system for tackling theselimitations, is centered around three techniques . It uses a single object detection model with multiple exit points for short-circuiting the inference .…

## An Implementation of Vector Quantization using the Genetic Algorithm Approach

Machine learning and genetic programming have produced promising results in many cases . The need for compression arises due to the exorbitant size of data shared on the internet . The paper alsodescribes an implementation of Vector Quantization using GA to generatecodebook which is used for Lossy image compression .…

## Arboreal Categories An Axiomatic Theory of Resources

Arboreal categories have intrinsic process structure, allowing dynamic notions such as bisimulation and back-and-forth games to be defined . These are related to extensional structures via arboreal covers, which areresource-indexed comonadic adjunctions . These ideas are developed in a verygeneral, axiomatic setting, and applied to relational structures .…

## Robust Factorization of Real world Tensor Streams with Patterns Missing Values and Outliers

Real-world tensor streams often include missing entries and unexpected outliers . SOFIA integrates them in linear time, in an online manner, despite the presence of missing entries . SoFIA is robust and accurate: yields up to 76% lower imputation error and 71% lower forecastingerror .…

## Orchestrated Trios Compiling for Efficient Communication in Quantum Programs with 3 Qubit Gates

Current quantum computers are especially error prone and require high level optimization to reduce operation counts and maximize the probability the program will succeed . We propose a new compiler structure that first decomposes to the three-qubitToffoli, routes the inputs of the higher-level Toffoli operations to groups of qubits, then finishes decomposition to hardware-supported gates .…

## Optimal intervention in traffic networks

We present an efficient algorithm to identify which edge should be improved in a traffic network to minimize the total travel time . We then study the optimality of the proposed procedure for recurrent networks, and provide simulations oversynthetic and real transportation networks .…

## About Weighted Random Sampling in Preferential Attachment Models

The Barabasi-Albert model is a very popular model for creating randomscale-free graphs . Despite its widespread use, there is a subtle ambiguity inthe definition of the model and, consequently, the dependent models and applications . This ambiguity is a result of a tight relation with the field of unequal probability random sampling, which dictates the exact process of edge creation after a newborn node has been added .…

## Compilation of mathematical expressions in Kotlin

KMath library uses the Kotlin object builder DSL and its own algebraic abstractions to generate an AST for mathematical operations . This AST is then compiled just-in-time to generate JVM bytecode . A similar approach is tested on other Kotlin platforms, where its performance is compared across avariety of supported platforms .…

## On Greedily Packing Anchored Rectangles

Freedman [1969] posed this problem in 1969, asking whether one can always cover at least 50% of U . Dumitrescuand T\’oth [2011] achieved the first constant coverage of 9.1% . Authors could not find any instance where their algorithm covers less than 50% .…

## Finite Atomized Semilattices

We show that every finite semilattice can be represented as an atomizedsemilattice . Each atom maps to one subdirectlyirreducible component, and the set of atoms forms a hypergraph that fully defines the semilatice . We also show that each atomization always exists and is unique up to “redundant atoms” The results can be applied to machine learning and to the study of semantic embeddings into algebras withidempotent operators .…

## Spectral formulation of the boundary integral equation method for antiplane problems

A spectral formulation of the boundary integral equation method for antiplaneproblems is presented . It involves evaluating a space-time convolution of the shearstress or the displacement discontinuity at the interface . In the spectralformulation, the convolution with respect to the spatial coordinate is performed in the spectral domain .…

## Interim envy freeness A new fairness concept for random allocations

We study thenovel notion of interim envy-freeness (iEF) for lotteries over allocations . iEF aims to serve as a sweet spot between the too stringent notion of ex-postenvy freeness and ex-ante envy-free . Our analysis relates iEF to other fairness notions as well, and reveals tradeoffs between iEF and efficiency .…

## Maximizing Conditional Entropy for Batch Mode Active Learning of Perceptual Metrics

Active metric learning is the problem of incrementally selecting batches of training data (typically, ordered triplets) to annotate, in order toprogressively improve a learned model of a metric over some input domain asrapidly as possible . Standard approaches, which independently select eachtriplet in a batch, are susceptible to highly correlated batches with many redundant triplets and hence low overall utility .…

## Roman Domination of the Comet Double Comet and Comb Graphs

The Roman dominationnumbers of the comet, double comet, and comb graphs are given in this paper . There are many kinds of dominating and relative types of sets ingraphs . We are going to focus on Roman domination, which is a type of domination that came up with an article by Ian Stewart .…

## RMIX Learning Risk Sensitive Policies for Cooperative Reinforcement Learning Agents

RMIX is a novel cooperative MARL method with the ConditionalValue at Risk (CVaR) measure over the learned distributions of individuals’ Q values . Empirically, our method significantly outperforms state-of-the-art methods on StarCraft II tasks, demonstrating enhanced coordination and improved sample efficiency .…

## Prioritizing Original News on Facebook

This work outlines how we prioritize original news, a critical indicator of quality . By examining the landscape and life-cycle of news posts on our social media platform, we identify challenges of building and deploying anoriginality score . We pursue an approach based on normalized PageRank values and three-step clustering, and refresh the score on an hourly basis to capture the dynamics of online news .…

## Short and long term prediction of a chaotic flow A physics constrained reservoir computing approach

We propose a physics-constrained machine learning method-based on reservoircomputing- to time-accurately predict extreme events and long-term velocitystatistics in a model of turbulent shear flow . We show that the combination of the two approaches is able to accurately reproduce the velocity statistics and topredict the occurrence and amplitude of extreme events .…

## Annealed Flow Transport Monte Carlo

Annealed Importance Sampling (AIS) and its Sequential Monte Carlo (SMC)extensions are state-of-the-art methods for estimating normalizing constants of probability distributions . We propose a novel Monte Carlo algorithm,Annealed Flow Transport (AFT) that builds upon AIS and SMC and combines them with normalizing flows (NF) for improved performance .…

## Distributed Online Learning for Joint Regret with Communication Constraints

In this paper we consider a distributed online learning setting for joint regret with communication constraints . This is amulti-agent setting in which in each round $t$ an adversary activates an agent, which has to issue a prediction . A subset of all the agents may then communicate a $b$-bit message to their neighbors in a graph .…