## Utilizing Redundancy in Cost Functions for Resilience in Distributed Optimization and Learning

This paper considers the problem of resilient distributed optimization andstochastic machine learning in a server-based architecture . We consider the case when some of the agents may beasynchronous and/or Byzantine faulty . In this case, the classical algorithm ofdistributed gradient descent (DGD) is rendered ineffective .…

## Neuro Symbolic Reinforcement Learning with First Order Logic

Deep reinforcement learning (RL) methods often require many trials before convergence, and no direct interpretability of trained policies is provided . We propose a novel RL method for text-based games with a recent neuro-symbolicframework called Logical Neural Network . The method is first toextract first-order logical facts from text observation and external wordmeaning network (ConceptNet) then train a policy in the network with directlyinterpretable logical operators .…

## A Fine Grained Analysis on Distribution Shift

Robustness to distribution shifts is critical for deploying machine learning models in the real world . Despite this necessity, there has been little work indefining the underlying mechanisms that cause these shifts . We find progress has been made over a standard ERMbaseline; in particular, pretraining and augmentations (learned or heuristic)offer large gains in many cases .…

## OpenABC D A Large Scale Dataset For Machine Learning Guided Integrated Circuit Synthesis

Logic synthesis is a challenging and widely-researched combinatorial optimization problem during integrated circuit (IC) design . There are no standarddatasets or prototypical learning tasks defined for this problem domain . Here, we describe OpenABC-D, a large-scale, labeled dataset produced by synthesizing open source designs with a leading open-source logic synthesis tool .…

## A Deep Insight into Measuring Face Image Utility with General and Face specific Image Quality Metrics

Paper focuses on face images and measurement of face image utility with general and face-specific image quality metrics . Quality scores provide a measure to evaluate the utility of biometric samples for biometric recognition . Individual handcrafted features lack general stability and perform significantly worse than general face specific quality metrics, the authors say .…

## A scale invariant ranking function for learning to rank a real world use case

Online Travel Agencies provide the main service for booking holidays, business trips, accommodations, etc. As in many e-commerce services, the use of a RecommenderSystem facilitates the navigation of the marketplaces . One of the main challenges when productizing machine learning models is the need of, not only consistent pre-processingtransformations, but also input features maintaining a similar scale both attraining and prediction time .…

## Weakly Supervised Training of Monocular 3D Object Detectors Using Wide Baseline Multi view Traffic Camera Data

Accurate 7DoF prediction of vehicles at an intersection is an important task . In principle, this could be achieved by a single camera system that is capable of detecting the pose of each vehicle but this would require a large, accurately labelled dataset fromwhich to train the detector .…

## CNewSum A Large scale Chinese News Summarization Dataset with Human annotated Adequacy and Deducibility Level

A large-scale Chinese newssummarization dataset CNewSum consists of 304,307 documents and human-written summaries for the news feed . It has long documents with high-abstractive summaries, which can encourage document-level understandingand generation for current summarization models . The test set contains adequacy and deducibilityannotations for the summaries .…

## Robust Edge Direct Visual Odometry based on CNN edge detection and Shi Tomasi corner optimization

In this paper, we propose a robust edge-direct visual odometry (VO) based on CNN edge detection and Shi-Tomasi corner optimization . Four layers of pyramid were extracted from the image in the proposed method to reduce the motion error between frames .…

## RoQNN Noise Aware Training for Robust Quantum Neural Networks

Quantum Neural Network (QNN) is a promising application towards quantumadvantage on near-term quantum hardware . The performance of QNN models has a severe degradation on realquantum devices . The accuracy gap between noise-free simulation and noisy results on IBMQ-Yorktown for MNIST-4 classification is over 60% .…

## Extraction of Positional Player Data from Broadcast Soccer Videos

Computer-aided support and analysis are becoming increasingly important inthe modern world of sports . The scouting of potential prospective prospective players, performance as well as match analysis, and the monitoring of training programsrely more and more on data-driven technologies . Many approaches require large amounts of data, which are, however, not easy to obtain in general .…

## Multimodal Learning using Optimal Transport for Sarcasm and Humor Detection

Multimodal learning is an emerging yet challenging research area . Being a fleeting action, sarcasm detection is challenging since large datasets are not available for this task in the literature . We propose a novel system, MuLOT, which utilizes self-attention to exploitintra-modal correspondence and optimal transport .…

## Super resolution of multiphase materials by combining complementary 2D and 3D image data using generative adversarial networks

Modelling the impact of a material’s mesostructure on device levelperformance typically requires access to 3D image data containing all therelevant information to define the geometry of the simulation domain . In this paper, we present a method for combining information from pairs of distinct but complementary imaging techniques in order to accurately reconstruct the desired multi-phase, high resolution,representative, 3D images .…

## On Hard Episodes in Meta Learning

Existing meta-learners primarily focus on improving the average task accuracy across multiple episodes . Different episodes may vary in hardness and quality leading to a wide gap in the meta-learner’s performance acrossepisodes . Understanding this issue is particularly critical in industrialfew-shot settings, where there is limited control over test episodes as they are uploaded by end-users .…

## Improving Non autoregressive Generation with Mixup Training

Pre-trained language models have achieved great success on various language understanding tasks, but how to effectively leverage them intonon-autoregressive generation tasks remains a challenge . To bridge the gap between autoregressive and non-autorgressive models, we propose a simple and effective trainingmethod called MIx Source and pseudo Target (MIST) Unlike other iterative training methods, MIST works in the training stage and has no effect on inference time .…

## StyleAlign Analysis and Applications of Aligned StyleGAN Models

In this paper, we perform an in-depth study of the properties and applications of aligned generative models . We refer to two models as aligned if they share the same architecture, and one of them (the child) is obtained from the other (the parent) via fine-tuning to another domain .…

## Mixer based lidar lane detection network and dataset for urban roads

Lidar point cloud has little image distortion in the BEV-projection . Lane Mixer Network (LMN) achieves the state-of-the-art performance, an F1 score of 91.67%, with K-Lane. We provide a world-first large urbanlane dataset for LIDar, which has maximum 6 lanes under various urbanroad conditions.…

## Learning 3D Semantic Segmentation with only 2D Image Supervision

With the recent growth of urban mapping and autonomous driving efforts, there has been an explosion of raw 3D data collected from terrestrial platforms withlidar scanners and color cameras . The proposed network architecture, 2D3DNet,achieves significantly better performance (+6.2-11.4 mIoU) than baselines during experiments on a new urban dataset with lidar and images captured in 20cities across 5 continents .…

## Data driven finite element method with RVE generated foam data

The data-driven finite element method proposed by Kirchdoerfer and Ortiz [1]allows to elude the material modeling step . Instead, a previously obtained dataset is used directly in the algorithm to describe the material behavior underdeformation . Usually, this data set is expected to be gained experimentally .…

## Topic Guided Abstractive Multi Document Summarization

A critical point of multi-document summarization (MDS) is to learn therelations among various documents . In this paper, we propose a novelabstractive MDS model, in which we represent multiple documents as aheterogeneous graph . We adopt amulti-task learning strategy to jointly train the topic and summarization module, allowing the promotion of each other .…

## On games and simulators as a platform for development of artificial intelligence for command and control

Games and simulators can be a valuable platform to execute complexmulti-agent, multiplayer, imperfect information scenarios with significantparallels to military applications . Multiple participants manage resources and make decisions that command assets to secure specific areas of a map or neutralize opposing forces .…

## Improving the Deployment of Recycling Classification through Efficient Hyper Parameter Analysis

The newly developed model scores a test-set accuracy of 95.8\% with a real word accuracy of . 95%, a14% increase over the original . Our acceleration pipeline boosted modelthroughput by 750% to 24 inferences per second on the Jetson Nano on the .…

## Actor critic is implicitly biased towards high entropy optimal policies

We show that the simplest actor-critic method — a linear softmax policyupdated with TD through interaction with a linear MDP — does not merely find an optimalpolicy, but prefers high entropy optimal policies . The algorithm not only has no regularization, noprojections, and no exploration like $\epsilon$-greedy, but is moreover trained on a single trajectory with no resets .…

## HCV Hierarchy Consistency Verification for Incremental Implicitly Refined Classification

Hierarchy-ConsistencyVerification (HCV) is an enhancement to existing continual learning methods . Our method incrementally discovers the hierarchical relations between classes . We then show how this knowledge can be exploited during both training andference . Code is available inhttps://://://github.com/wangkai930418/HCV_IIRC and HCV_Hierarchy-ConferenceVerification is available to download in dro dro droplets of code .…

## Each Attribute Matters Contrastive Attention for Sentence based Image Editing

Sentence-based Image Editing (SIE) aims to deploy natural language to edit animage . But existing methods can hardly produce accurate editing and even lead to failures in attribute editing when the querysentence is with multiple editable attributes . To cope with this problem, this paper proposes anovel model called Contrastive Attention Generative Adversarial Network (CA-GAN) which is inspired from contrastive training .…

## Fast Model Editing at Scale

Model Editor Networks with Gradient Decomposition (MEND) is a collection of small auxiliary editing networks that use a single desired input-output pair to make fast, local edits to a pre-trained model . MEND learns to transform the gradient obtained by standard fine-tuning, using a low-rank decomposition of the gradient to make the parameterization of this transformation tractable .…

## 3D ANAS v2 Grafting Transformer Module on Automatically Designed ConvNet for Hyperspectral Image Classification

Hyperspectral image (HSI) classification has been a hot topic for decides, as HSI classification has rich spatial and spectral information, providing strongbasis for distinguishing different land-cover objects . Recently, severalneural architecture search (NAS) algorithms are proposed for HSI classifications . We propose a novel hybrid search space, where 3Dconvolution, 2D spatial convolution and 2D spectral convolution are employed .…

## LOA Logical Optimal Actions for Text based Interaction Games

Logical Optimal Actions (LOA) is an action decision architecture ofreinforcement learning applications with a neuro-symbolic framework . LOA is a combination of neural network and symbolic knowledge acquisition approach for natural language interaction games . The demonstration for LOA experimentsconsists of a web-based interactive platform for text-based games and visualization for acquired knowledge for improving interpretability for trained rules .…

## A Secretive Coded Caching for Shared Cache Systems using PDAs

This paper considers the secretive coded caching problem with shared caches . In ashared cache network, the users are served by a smaller number of helper caches . Each user is connected to exactly one helper cache . To ensure the secrecyconstraint in shared cache networks, each user is required to have an individual cache of at least unit file size .…

## Single Modal Entropy based Active Learning for Visual Question Answering

Constructing a large-scale labeled dataset in the real world, especially for high-level tasks (eg, Visual Question Answering), can be expensive and time-consuming . We propose a novel method for effectivesample acquisition through the use of ad hoc single-modal branches for each input to leverage its information .…

## Dual Encoding U Net for Spatio Temporal Domain Shift Frame Prediction

The landscape of city-wide mobility behaviour has altered significantly over the past 18 months . The ability to make accurate and reliable predictions on such behaviour has likewise changed drastically . This paper seeks to address this question by introducing an approach for traffic frame prediction using a lightweight dual-Encoding U-Net built using only 12 Convolutional layers .…

## User friendly introduction to PAC Bayes bounds

PAC-Bayesian orPAC-Bayes bounds are a set of tools designed to understand the generalization ability of such procedures . Aggregated and randomized predictors are not defined by a minimization problem, but by a probability distribution on the set of predictors . An elementary introduction to PAC- Bayesian theory is still missing .…

## Multiobjective Dijkstra A

We introduce the Multiobjective Dijkstra A* (MDA*) algorithm for the One-to-One MultiObjective Shortest Path Problem . The algorithm requires a monotone node heuristic as part of its input . For any node, the heuristic underestimates the costs of apath from this node to the target node of the search .…

## Analyzing and Improving the Optimization Landscape of Noise Contrastive Estimation

Noise-contrastive estimation (NCE) is a statistically consistent method for learning unnormalized probabilistic models . It has been empirically observed that the choice of the noise distribution is crucial for NCE’s performance . In this work, we prove these challengesarise due to an ill-behaved (more precisely, flat) loss landscape .…

## OpenABC D A Large Scale Dataset For Machine Learning Guided Integrated Circuit Synthesis

Logic synthesis is a challenging and widely-researched combinatorial optimization problem during integrated circuit (IC) design . There are no standarddatasets or prototypical learning tasks defined for this problem domain . Here, we describe OpenABC-D, a large-scale, labeled dataset produced by synthesizing open source designs with a leading open-source logic synthesis tool .…

## FedGEMS Federated Learning of Larger Server Models via Selective Knowledge Fusion

Federated Learning (FL) hasemerged as a viable solution to learn a global model while keeping dataprivate . But model complexity of FL is impeded by the computation resources of edge nodes . In this work, we investigate a novel paradigm to take advantage of a powerful server model to break through model capacity in FL .…

## Deep Curriculum Learning in Task Space for Multi Class Based Mammography Diagnosis

Deeplearning techniques have succeeded in reaching near-human performance in anumber of tasks . Mammography is used as a standard screening procedure for the potential patients of breast cancer . We propose an end-to-endCurriculum Learning (CL) strategy in task space for classifying the three categories of Full-Field Digital Mammography (FFDM), namely Malignant,Negative, and False recall .…

## Each Attribute Matters Contrastive Attention for Sentence based Image Editing

Sentence-based Image Editing (SIE) aims to deploy natural language to edit animage . But existing methods can hardly produce accurate editing and even lead to failures in attribute editing when the querysentence is with multiple editable attributes . To cope with this problem, this paper proposes anovel model called Contrastive Attention Generative Adversarial Network (CA-GAN) which is inspired from contrastive training .…

## Learning OFDM Waveforms with PAPR and ACLR Constraints

Most modern systems useorthogonal frequency-division multiplexing (OFDM) for its efficient equalization . This waveform suffers from multiple limitations such as a highadjacent channel leakage ratio (ACLR) and high peak-to-average power ratio (PAPR) In this paper, we propose a learning-based method to design OFDM-basedwaveforms that satisfy selected constraints while maximizing an achievableinformation rate .…

## Evolutionary Foundation for Heterogeneity in Risk Aversion

We examine the evolutionary basis for risk aversion with respect to aggregaterisk . We study populations in which agents face choices between aggregate riskand idiosyncratic risk . We show that the choices that maximize the long-rungrowth rate are induced by a heterogeneous population in which the least and most risk averse agents are indifferent between aggregate and obtaining its linear and harmonic mean for sure .…

## Reinforcement Learning Based Optimal Camera Placement for Depth Observation of Indoor Scenes

An OCP online solution to depth observation of indoor scenes based on reinforcementlearning is proposed in this paper . The proposed system outperforms seven out often test scenes in obtaining lower depth observation error . The total error inall test scenes is also less than 90% of the baseline ones.…

## Variational Predictive Routing with Nested Subjective Timescales

Variational Predictive Routing is a neural probabilistic inference system that organizes latent representations of video features in a temporal hierarchy, based on their rates of change . VPR is able to detect event boundaries, disentangle spatiotemporalfeatures across its hierarchy, adapt to the dynamics of the data, and produce accurate time-agnostic rollouts of the future .…

## Autonomous Dimension Reduction by Flattening Deformation of Data Manifold under an Intrinsic Deforming Field

A new dimension reduction (DR) method for data sets is proposed by autonomousdeforming of data manifolds . The deformation is guided by the proposeddeforming vector field, defined by two kinds of virtual interactions between data points . The flattening of data manifold is achieved as an emergentbehavior under the elastic and repelling interactions .…

## Dual Encoding U Net for Spatio Temporal Domain Shift Frame Prediction

The landscape of city-wide mobility behaviour has altered significantly over the past 18 months . The ability to make accurate and reliable predictions on such behaviour has likewise changed drastically . This paper seeks to address this question by introducing an approach for traffic frame prediction using a lightweight dual-Encoding U-Net built using only 12 Convolutional layers .…

## Generalized Out of Distribution Detection A Survey

Out-of-distribution (OOD) detection is critical to ensuring the reliabilityand safety of machine learning systems . Other problems related to OOD detection include anomaly detection (AD), novelty detection (ND), open set recognition (OSR), and outlier detection (OD) Despite having different definitions and settings, these problems often confuse readers and practitioners, some existing studies misuse terms .…

## The popular assignment problem when cardinality is more important than popularity

We consider a matching problem in a bipartite graph $G=(A\cup B,E)$ whereeach node in $A$ is an agent having preferences in partial order over herneighbors, while nodes in $B$ are objects with no preferences . The goal is to compute an assignment $M$ such that there is no better or “more popular” assignment .…

## Algorithmic Amplification of Politics on Twitter

Content on Twitter’s home timeline is selected and ordered by personalizational algorithms . By consistently ranking certain content higher, these algorithmsmay amplify some messages while reducing the visibility of others . In 6 out of 7 countries studied, the mainstream political right enjoys higher algorithmic amplification than themainstream political left .…

## Inverse Optimal Control Adapted to the Noise Characteristics of the Human Sensorimotor System

Computational level explanations based on optimal feedback control with signal-dependent noise have been able to account for a vast array of phenomenain human sensorimotor behavior . We formalize the problem as apartially observable Markov decision process and distinguish between theagent’s and the experimenter’s inference problems .…

## Reinforcement Learning Based Optimal Camera Placement for Depth Observation of Indoor Scenes

An OCP online solution to depth observation of indoor scenes based on reinforcementlearning is proposed in this paper . The proposed system outperforms seven out often test scenes in obtaining lower depth observation error . The total error inall test scenes is also less than 90% of the baseline ones.…

## Generalized Out of Distribution Detection A Survey

Out-of-distribution (OOD) detection is critical to ensuring the reliabilityand safety of machine learning systems . Other problems related to OOD detection include anomaly detection (AD), novelty detection (ND), open set recognition (OSR), and outlier detection (OD) Despite having different definitions and settings, these problems often confuse readers and practitioners, some existing studies misuse terms .…