The AI Arena A Framework for Distributed Multi Agent Reinforcement Learning

Advances in reinforcement learning (RL) have resulted in recent breakthroughs in the application of artificial intelligence (AI) We introduce the AI Arena: a scalable framework with flexible abstractions for distributedmulti-agent reinforcement learning . The AI Arena extends the OpenAI Gyminterface to allow greater flexibility in learning control policies across multiple agents with heterogeneous learning strategies and localized views of the environment .…

A Training Set Subsampling Strategy for the Reduced Basis Method

We present a subsampling strategy for the offline stage of the Reduced Basis Method . The approach is aimed at bringing down the considerable offline costs associated with using a finely-sampled training set . The proposed algorithmexploits the potential of the pivoted QR decomposition and the discrete empirical interpolation method to identify important parameter samples .…

Multi Objective Resource Allocation for IRS Aided SWIPT

In this letter, we study the resource allocation for a multiuser intelligentreflecting surface (IRS)-aided simultaneous wireless information and powertransfer (SWIPT) system . A multi-antenna base station (BS)transmits energy and information signals simultaneously to multiple energyharvesting receivers (EHRs) and information decoding receivers (IDRs) assisted by an IRS .…

Semantics Consistent Representation Learning for Remote Sensing Image Voice Retrieval

Cross-modal RS image-voice retrieval provides a new insight . The proposed SCRL method consists of two main steps: 1) semantics encoding and 2)semantics-consistent representation learning . The main novelty is that the proposed method takes the pairwise, intra-modality, and non-paired intermodality relationships into account simultaneously, thereby improving the semantic consistency of the learned representations for the RS image .…

Novel tile segmentation scheme for omnidirectional video

Regular omnidirectional video encoding technics use map projection to flattena scene from a spherical shape into one or several 2D shapes . Common projectionmethods including equirectangular and cubic projection have varying levels ofpolation that create a large number of non-information-carrying pixelsthat lead to wasted bitrate .…

Spatial Attention based Non reference Perceptual Quality Prediction Network for Omnidirectional Images

A spatialattention-based perceptual quality prediction network for non-reference qualityassessment on ODIs (SAP-net) The network can adaptivelyestimate human perceptual quality on impaired ODIs through a self-attentionmanner, which significantly promotes the prediction performance of quality scores . Extensive experiments validate that our network outperforms 9 state-of-the-art methods for quality assessment on ODI .…

Machine Learning for Massive Industrial Internet of Things

Industrial Internet of Things (IIoT) revolutionizes the future manufacturingfacilities . With the deployment of massive IIoT devices, it is difficult for the wireless network to support the ubiquitous connections with diversequality-of-service (QoS) requirements . Machine learning is regarded as a powerful data-driven tool to optimize wireless network .…

Quantum Algorithms in Cybernetics

A new method for simulation of a binary homogeneous Markov process using aquantum computer was proposed . The IBM’sYorktown ‘bow-tie’ back-end performs quite better than the 5-qubit T-shaped and 14-quinst Melbourne quantum processors in terms of quantum fidelity . The creation of a quantum controlled n-throot of X gate using only the existing basic quantum logic gates at the available cloud platforms is possible, although the hardware devices are still too noisy, which results in a significant measurement error increase.…

An exponential integrator WENO discretization for sonic boom simulation on modern computer hardware

Splitting methods allow one to divide complicatedpartial differential equations into simpler parts that are solved by tailored numerical schemes . The new approach also improves the accuracy compared to the splitting method and it reduces oscillations. The numericalmethod is designed to be highly parallelisable and therefore takes full advantage of modern computer hardware.…

EBP GEXIT Charts for M ary AWGN Channel for Generalized LDPC and Turbo Codes

This work proposes a tractable estimation of the maximum a posteriori (MAP)threshold of various families of sparse-graph code ensembles . We estimate the MAP threshold by applying the Maxwellconstruction to the obtained approximate EBP-GEXIT charts . We also explore where the spatially-coupledbelief propagation (BP) threshold is located with respect to the previouslycomputed MAP threshold .…

Towards a Cognitive Compute Continuum An Architecture for Ad Hoc Self Managed Swarms

In this paper we introduce our vision of a Cognitive Computing Continuum to address the changing IT service provisioning towards a distributed,opportunistic, self-managed collaboration between heterogeneous devices . The focal point of this continuum are cognitive devices, which have to make decisions autonomously using their on-board computation and storage capacity based on information sensed from their environment .…

Learning to compose 6 DoF omnidirectional videos using multi sphere images

The system uses a 3D ConvNet to generate a multi-sphere images (MSI)representation that can be experienced in 6-DoF VR . The system utilizes conventional VR camera footage directly without the need for adepth map or segmentation mask . By using a newlydesigned weighted sphere sweep volume (WSSV) fusing technique, our approach is compatible with most panoramic VR camera setups .…

Evaluating Linear Functions to Symmetric Monoidal Categories

Symmetric Monoidal Categories (SMCs) are pervasive algebraic structures . We provide a method to evaluate linear functions to their own representation as a morphism in a SMC . We believe this method is well-suited as a basis for embedding domain specific languages whose computational model can be framed as an instance of SMC.…

Unified Pre training for Program Understanding and Generation

PLBART is pre-trained on an extensive collection of Java and Python functions and associated NL text viadenoising autoencoding . The model outperforms or rivals state-of-the-art models . Analysis reveals that PLBart learns program syntax,style (e.g., identifier naming convention) and logical flow that are crucial to programsemantics and thus excels even with limited annotations .…

What Have We Learned from OpenReview

Anonymous peer review is used by the great majority of computer science conferences . The paper, (meta) reviews, rebuttals, and final decisions are all released to public . We collect 5,527 submissions and their 16,853reviews from the OpenReview platform . We also collect these submissions’ citation data from Google Scholar and their non-peer-reviewed versions fromarXiv.org…

Generalized continuation Newton methods and the trust region updating strategy for the underdetermined system

Paper considers the generalized continuation Newton method and thetrust-region updating strategy for the underdetermined system of nonlinearequations . New method uses a switching updating technique of the Jacobian matrix to improve its computational efficiency . The computational speed of thenew method is about eight to fifty times as fast as that of fsolve.m…

Bounded Invariant Checking for Stateflow Programs

Stateflow models are complex software models, often used as part of safety-critical software solutions designed with Matlab Simulink . They incorporate design principles that are typically very hard to verify formally . The Stateflow language lacks a formal semantics, which hinders the formal analysis .…

Learning of Causal Observable Functions for Koopman DFL Lifting Linearization of Nonlinear Controlled Systems and Its Application to Excavation Automation

Koopman operator theory allows us to represent a nonlinear system as alinear system in an infinite-dimensional space of observables . But exactlinearization is guaranteed only for autonomous systems with no input, and finding effective observable functions for approximation with a low-order linear system remains an open question .…

A principled approach for weighted multilayer network aggregation

A multilayer network depicts different types of interactions among the sameset of nodes . Different layers represent distinct types of experimentally confirmedmolecule interactions among proteins . In some researches theco-expression layer is in turn represented as a multilayered network . We propose a maximum a posteriori estimation-based algorithm .…

Congestion control in high speed networks using the probabilistic estimation approach

The bulk of Internet traffic uses TCP protocol for reliable transmission . But the standard TCP’s performance is very poor in High Speed Networks . This problem has roots in conservative nature of TCP, especially in its AIMD phase . To this end, this paper proposes analgorithm which considers packet loss and delay information jointly and employsa probabilistic approach to accurately estimation of congestion status in thenetwork .…

Search Disaster Victims using Sound Source Localization

Sound Source Localization (SSL) are used to estimate the position of soundsources . Various methods have been used for detecting sound and itslocalization . This paper presents a system for stationary sound sourcelocalization by cubical microphone array . The computed azimuth and elevation angle of victimizedhuman voice are fed to embedded omni-directional drive system which navigates vehicle automatically towards the stationary sound sources .…

Spheroidal Ambisonics a Spatial Audio Framework Using Spheroidal Bases

Ambisonics is an established framework to capture, process, and reproducespatial sound fields based on its spherical harmonics representation . Wepropose a generalization of conventional spherical ambisonics to the spheroidalcoordinate system . This framework is referred to asspheroidal ambisonic and a formulation for the case of prolate spheoidalcoordinates is presented .…

Developing and evaluating an human automation shared control takeover strategy based on Human in the loop driving simulation

A “human-in-the-loop” driving simulator experiment was conducted to evaluate the impact of the proposed shared control takeover strategy under different disengagement conditions . Results of thirty-two drivers showed sharedcontrol takeover strategy could improve safety performance at the aggregatedlevel . For more urgentdisengagements caused by another vehicle’s sudden brake, a shared controlstrategy enlarges individual differences .…

Repairing Serializability Bugs in Distributed Database Programs via Automated Schema Refactoring

Serializability is a well-understood concurrency control mechanism that easesreasoning about highly-concurrent database programs . Unfortunately, enforcingserializability has a high-performance cost, especially on geographically distributed database clusters . We present a sound andfully automated refactoring procedure that transforms a program’s datalayout to eliminate concurrency bugs .…

Graph Metrics for Internet Robustness A Survey

Research on the robustness of the Internet has gained critical importance inthe last decades because more and more individuals, societies and firms rely on this global network infrastructure for communication, knowledge transfer, business processes and e-commerce . The Internet has inspired several novel graph metrics for assessing important topological robustness features of large complex networks such as the Internet .…

Linear Mapping based Variational Ensemble Kalman Filter

We propose a linear-mapping based variational Ensemble Kalman filter for problems with generic observation models . The proposed method isformulated as to construct a linear mapping from the prior ensemble to theposterior one . The linear mapping is computed by minimizing theKullback-Leibler divergence between the transformed distribution and the actual posterior .…

BROOD Bilevel and Robust Optimization and Outlier Detection for Efficient Tuning of High Energy Physics Event Generators

Monte Carlo event generators are tuned on experimentalmeasurements by evaluating the goodness of fit between the data and the MCpredictions . The relative importance of each measurement is adjusted manually to meet different experimental needs . We introduce several optimization formulations and algorithms with new criteria for streamlining and automating this process .…

Analyzing Human Models that Adapt Online

Predictive human models often need to adapt their parameters online from human data . This raises previously ignored safety-related questions for robots relying on these models . Key idea is to model therobot’s learning algorithm as a dynamical system where the state is the currentmodel parameter estimate and the control is the human data the robot observes .…

Fast and flexible Human program induction in abstract reasoning tasks

The Abstraction and Reasoning Corpus (ARC) is a challenging program inductiondataset that was recently proposed by Chollet (2019) Here, we report the first set of results collected from a behavioral study of humans solving a subset oftasks from ARC (40 out of 1000) Although this subset of tasks contains considerable variation, our results showed that humans were able to infer theunderlying program and generate the correct test output for a novel test input .…

The state dependence of integrated instantaneous and fluctuating entropy production in quantum and classical processes

Given a fixed physical process, we consider the entropy production (EP)incurred by some initial state $rho$ additional to the minimal EP incurred by the least-dissipative state $varphi$ We show that this additional EP has a universal information-theoretic form, given by contraction of the relativeentropy between the initial state and the least in time .…

Effectively Counting s t Simple Paths in Directed Graphs

An important tool in analyzing complex social and information networks is s-tsimple path counting, which is known to be #P-complete . In this paper, we studyefficient s-t simple path counting in directed graphs . For a given pair ofvertices s and t in a directed graph, first we propose a pruning technique that can efficiently and considerably reduce the search space .…

Mixed Reality Interaction Techniques

This chapter gives an overview of interaction techniques for mixed reality including augmented and virtual reality (AR/VR) Various modalities for input and output are discussed . Techniques for tangible and surface-based interaction include gesture-based, pen-based and gaze-based . Interaction with intelligent virtual agents is considered in the book .…

OpenTera A Microservice Architecture Solution for Rapid Prototyping of Robotic Solutions to COVID 19 Challenges in Care Facilities

Telehealth frameworks are becoming more widely adopted in the context of long-term care (LTC) for older adults . This paper presents OpenTera, an open source telehealth framework, to facilitate prototyping of such solutions by software and roboticdesigners . The need to deal with the effects of the COVID-19 pandemic emphasizes the benefits of this approach, but also highlights new challenges for which robots could be interesting solutions to bedeployed in LTC facilities .…