## Unsupervised Learning of Monocular Depth and Ego Motion Using Multiple Masks

A new unsupervised learning method of depth and ego-motion using multiplemasks from monocular video is proposed in this paper . The method is to use a geometricrelationship to filter the mismatched pixels for training . The experiments on KITTI dataset show our method achieves good performance in terms of depth .…

## Design and development of an Aerial Surveillance Security System

Aerial security means performing security-aimed monitoring and surveillanceoperations with the help of airborne vehicles . Human officers (security organizations, law enforcement, police etc.) would be able to remotely monitor and view video and data acquired from Drones while planning and evaluating their operations .…

## Using multidimensional speckle dynamics for high speed large scale parallel photonic computing

The recent rapid increase in demand for data processing has resulted in the need for novel machine learning concepts and hardware . Physical reservoircomputing and an extreme learning machine are novel computing paradigms basedon physical systems themselves . The speckle-based mapping of the input information is high-dimensional and nonlinear and can berealized at the speed of light; thus, nonlinear time-dependent informationprocessing can successfully be achieved at fast rates when applying areservoir-computing-like-approach .…

## Cooperative UWB Based Localization for Outdoors Positioning and Navigation of UAVs aided by Ground Robots

Unmanned aerial vehicles (UAVs) are becoming largely ubiquitous with anincreasing demand for aerial data . Accurate navigation and localization often relies on RTK GNSS . Inexpensive ultra-wideband (UWB) transceivers enable centimeter-level relative positioning . With fast deployment and wide setup flexibility, the proposed system is able to accommodate different environments and can also beutilized in GNSS-denied environments .…

## Classically Verifiable Quantum Advantage from a Computational Bell Test

We propose and analyze a novel interactive protocol for demonstrating quantumcomputational advantage . Ourprotocol relies upon the cryptographic hardness of trapdoor claw-free functions . Through a surprising connection to Bell’s inequality, our protocolavoids the need for an adaptive hardcore bit, with essentially no increase inthe quantum circuit complexity and no extra cryptographic assumptions .…

## Cortical Morphometry Analysis based on Worst Transportation Theory

Biomarkers play an important role in early detection and intervention in Alzheimer’s disease (AD) However, obtaining effective biomarkers for AD is still a big challenge . The worst transportation (WT) aims to find the least economical way to transport one measure to the other, which contrasts to the optimal (OT) The WT map is the gradient of a concave function satisfying the Monge-Ampere equation .…

## Towards Evaluating and Training Verifiably Robust Neural Networks

CROWN, a bounding method based ontight linear relaxation, often gives very loose bounds on these networks . We also design a new activation function, parameterized ramp function (ParamRamp) which has more diversity of neuron status than ReLU . We conduct extensive experiments onMNIST, CIFAR-10 and Tiny-ImageNet with ParamRamp activation and achievestate-of-the-art verified robustness.…

## Positive Sample Propagation along the Audio Visual Event Line

Visual and audio signals often coexist in natural environments, forming audio-visual events (AVEs) Given a video, we aim to localize video segments containing an AVE and identify its category . In order to learn discriminativefeatures for a classifier, it is pivotal to identify the helpful (or positive)audio-visual segment pairs while filtering out the irrelevant ones .…

## Replicate or Relocate Non Uniform Access in Parameter Servers

Parameter servers (PSs) facilitate the implementation of distributed trainingfor large machine learning tasks . Parameter access is non-uniform in many real-world machine-learning tasks . Skew and nondeterminism are two major sources for non-Uniformity . Lapse2 outperformed existing, single-technique PSs by up to one order of magnitude .…

## Neural Video Portrait Relighting in Real time via Consistency Modeling

Video portraits relighting is critical in user-facing human photography, especially for immersive VR/AR experience . Recent advances still fail to cover consistent relit result under dynamic illuminations from monocular RGBstream, suffering from the lack of video consistency supervision . In thispaper, we propose a neural approach for real-time, high-quality and coherentvideo portrait relighting, which jointly models the semantic, temporal andlighting consistency .…

## Multi rate attention architecture for fast streamable Text to speech spectrum modeling

High-quality spectrum models usually incorporate the encoder-decoder architecture with self-attention orbi-directional long short-term (BLSTM) units . While these models can produce high quality speech, they often incur O($L$) increase in both latency and RTF with respect to input length $L$. Long input leads to longer delay and slower synthesis speed, limiting its use in real-time applications .…

## An Energy Efficient Quad Camera Visual System for Autonomous Machines on FPGA Platform

The visual frontend is a major performance and energy consumption bottleneck in autonomous machine applications . Compared to Nvidia TX1 and Intel i7, ourFPGA-based implementation achieves 5.6x and 3.4x speedup, as well as 3.0x and 34.6X power reduction, respectively. Compared to the Nvidia TX-1, Intel i.7, the implementation achieves 3.5x power reduction .…

## Optimizer Fusion Efficient Training with Better Locality and Parallelism

Machine learning frameworks adopt iterative optimizers to train neuralnetworks . By reordering the forward computation, gradientcalculation, and parameter updating, our proposed method improves theefficiency of iterativeoptimizers . Experimental results demonstrate that we achieve an up to 20% training time reduction on various configurations .…

## E Commerce in Turkey and SAP Integrated E Commerce System

E-commerce is becoming an indispensable method with the increase of internet usage . SAP is a pioneer and leader in the company resource planning software sector . The SAP is very important forlarge-scale companies. They manage all their processes on SAP and itsintegration is important with other related software.…

## Sub GMN The Subgraph Matching Network Model

Subgraph matching is acrucial task in many fields, ranging from information retrieval, computervision, biology, chemistry and natural language processing . Yet subgraphmatching problem remains to be an NP-complete problem . Study proposes anend-to-end learning-based approximate method for subgraph matching task, calledsubgraph matching network (Sub-GMN) The proposed Sub-GMn firstly uses graphrepresentation learning to map nodes to node-level embedding .…

## Hereditary rigidity separation and density In memory of Professor I G Rosenberg

We observe that on aset $V$ with $m$ elements, there is a hereditarily rigid set made of $n$ tournaments . We ask if the sameinequality holds when the tournaments are replaced by linear orders . We show that $h_{\rm Lin}(m)$ is the least cardinal $n such that$m(m-1) and $d(V) is the topological density of the set of linear orders on$V) We do not know whether these equalities hold without any set theoretical hypothesis .…

## Efficient Set Based Approaches for the Reliable Computation of Robot Capabilities

To reliably model real robot characteristics, interval linear systems ofequations allow to describe families of problems that consider sets of values . This allows to easily account for typical complexities such as sets of jointstates and design parameters uncertainties . For eachclass, reliable and efficient polytope, n-cube, and n-ball inner approximations are presented .…

## The best laid plans or lack thereof Security decision making of different stakeholder groups

Cyber security requirements are influenced by the priorities and decisions of a range of stakeholders . No group of experts makes significantly better gamedecisions than anyone else, and that their biases lead them to not fullycomprehend what they are defending or how the defenses work .…

## DVMark A Deep Multiscale Framework for Video Watermarking

Video watermarking embeds a message into a cover video in an imperceptible manner . The message can be retrieved even if the video undergoes certain modifications or distortions . The new model consists of a novel multiscale design where the watermarks are distributed across multiple spatial-temporal scales .…

## Hetero functional Network Minimum Cost Flow Optimization A Hydrogen Natural Gas Network Example

This work aims to develop an optimization program for a dynamic, hetero-functional graphtheory-based model of an engineering system . The optimization program is demonstrated through the application of the program to a hydrogen-naturalgas infrastructure test case . Four distinct scenarios are optimized todemonstrate potential synergies or cascading network effects of policy acrossinfrastructures .…

## Intuitive Tasks Planning Using Visuo Tactile Perception for Human Robot Cooperation

Designing robotic tasks for co-manipulation necessitates to exploit not onlypriprioceptive but also exteroceptive information for improved safety andautonomy . Research proposes to formulateintuitive robotic tasks following human viewpoint by incorporatingvisuo-tactile perception . The visual data using depth cameras surveils and determines the object dimensions and human intentions while the tactile sensing ensures to maintain the desired contact to avoid slippage .…

## The k Colorable Unit Disk Cover Problem

In this article, we consider colorable variations of the Unit Disk Cover (CDUDC) problem . We propose a 4-approximation algorithm in $O(m^{7k)n\log k) time for this problem, where$k$is a positive integer . We also extend our algorithm to solve the .it$k\$-Colorable…

## O 1 Steiner Point Removal in Series Parallel Graphs

We study how to vertical-sparsify a graph while preserving both the graph’s metric and structure . The main engine of our approach is a newmetric decomposition for series-parallel graphs . Roughly, a hammock decomposition is a forest-like structure thatpreserves certain critical parts of the metric induced by a series parallelgraph .…

## A Survey on Natural Language Video Localization

Natural language video localization (NLVL) aims to locate a targetmoment from a video that semantically corresponds to a text query . In this paper, we present acomprehensive survey of the NLVL algorithms . We categorize them into supervised andweakly-supervised methods, following by the analysis of the strengths andweaknesses of each kind of methods .…

## Sample efficient Gear ratio Optimization for Biomechanical Energy Harvester

The biomechanical energy harvester is expected to harvest the electricenergies from human motions . A tradeoff between harvesting energy and keeping the user’s natural movements should be balanced via optimization techniques . CVT could continuously adjust its gear ratio to balance the tradeoff foreach task .…

## AdaPool A Diurnal Adaptive Fleet Management Framework using Model Free Deep Reinforcement Learning and Change Point Detection

Deep Reinforcement Learning (RL) suffers from catastrophicforgetting due to being agnostic to the timescale of changes in the distribution of experiences . This paper introduces an adaptive model-free deep reinforcement approach that can recognize and adapt to the diurnal patterns in the ride-sharing environment with car-pooling .…

## Optimization Algorithm for Feedback and Feedforward Policies towards Robot Control Robust to Sensing Failures

Model-free or learning-based control, in particular, reinforcement learning(RL), is expected to be applied for complex robotic tasks . Traditional RL requires a policy to be optimized is state-dependent, that means, the policy is a kind of feedback (FB) controllers . To be improved, RL can be improvedby dealing with the FB/FF policies, but to the best of our knowledge, amethodology for learning them in a unified manner has not been developed .…

## A Joint Network for Grasp Detection Conditioned on Natural Language Commands

Command Grasping Network(CGNet) proposes a model to directly output command satisficinggrasps from RGB image and textual command inputs . CGNet outperforms a cascaded object-retrieval and grasp detection baseline by alarge margin . Three physical experiments demonstrate the functionality andperformance of CGNet .…

## Touch based Curiosity for Sparse Reward Tasks

Touch-based Curiosity (ToC) learns what visibleobjects interactions are supposed to “feel” like . We encourage exploration by rewarding interactions where the expectation and the experience don’t match . We compare our cross-modal approach to single-modality (touch- or vision-only) approaches as well as othercuriosity-based methods and find that our method performs better and is moresample-efficient .…

## Residual Model Learning for Microrobot Control

A majority of microrobots are constructed using compliant materials that are difficult to model analytically, limiting the utility of traditional model-based controllers . We propose anovel framework residual model learning (RML) that leverages approximate modelsto substantially reduce the sample complexity associated with learning anaccurate robot model .…

## Qualitative Planning in Imperfect Information Games with Active Sensing and Reactive Sensor Attacks Cost of Unawareness

We consider the probabilistic planning problem where the agent (called Player1, or P1) can jointly plan the control actions and sensor queries in a sensornetwork . We model such an adversarial interaction using a formal model — areachability game with partially controllable observation functions .…

## Putting NeRF on a Diet Semantically Consistent Few Shot View Synthesis

We present DietNeRF, a 3D neural scene representation estimated from a fewimages . NeRF learns a continuous volumetric representation of a scene through multi-view consistency . We introduce an auxiliary semantic consistency loss that encourages realistic renderings at novel poses .…

## Trajectory Tracking of Underactuated Sea Vessels With Uncertain Dynamics An Integral Reinforcement Learning Approach

Underactuated systems like sea vessels have degrees of motion that are insufficiently matched by a set of independent actuation forces . An online machine learning mechanism based on integral reinforcement learning is proposed to find a solution for a class of nonlinear tracking problems with partial prior knowledge of the system dynamics .…

## Seeing through a Black Box Toward High Quality Terahertz TomographicImaging via Multi Scale Spatio Spectral Image Fusion

Strong water absorption nature and low noise tolerance lead to undesiredblurring and distortion of reconstructed terahertz images . MS3-Unet uses multi-scale branches to extract spatio-spectral features then processed by element-wise adaptive filters, and then fused to achieve high-quality image restoration .…

## Modeling High order Interactions across Multi interests for Micro video Reommendation

Self-over-CoAttention module uses co-attention to model correlation patterns across different levels . We propose a Self-Over-Coattention module to enhance user’s interest representation . Experimental results on filtered public datasets verify that our module is useful . We use self-attraction to model correlations patterns within a specificlevel of interest in micro-videos .…

## Drug Discovery Approaches using Quantum Machine Learning

Traditional drug discovery pipeline takes several years and cost billions of dollars . Classical machines cannot efficiently produce atypical patterns of quantum computers which might improve the training quality of learning tasks . We propose a suite of quantum machine learning techniques e.g.,generative…

## Distributed Video Adaptive Block Compressive Sensing

Video block compressive sensing has been studied for use in resource-strstrained scenarios, such as wireless sensor networks, but the approach still suffers from low performance and long reconstruction time . We propose two algorithms that leverage convolutional neuralnetwork components to reconstruct video with greatly reduced reconstructiontime .…

## PhySG Inverse Rendering with Spherical Gaussians for Physics based Material Editing and Relighting

PhySG is an end-to-end inverse rendering pipeline that includes afully differentiable renderer and can reconstruct geometry, materials, andillumination from scratch . Our frameworkrepresents specular BRDFs and environmental illumination using mixtures ofspherical Gaussians . We demonstrate, with both synthetic and real data, that our reconstructions not only enable rendering of novel viewpoints, but also physics-based appearance editing of materials and illumination .…

## Fusing RGBD Tracking and Segmentation Tree Sampling for Multi Hypothesis Volumetric Segmentation

The key challenge is estimating the segment boundaries of (partially) occluded objects, which areinherently ambiguous when considering only a single frame . We propose Multihypothesis Segmentation Tracking (MST), a novel method forvolumetric segmentation in changing scenes . MST outperforms baselines in all tested scenes, showing it outperforms baselines in all tests .…

## TL DR Out of Context Adversarial Text Summarization and Hashtag Recommendation

This paper presents Out-of-Context Summarizer, a tool that takes arbitrarypublic news articles out of context by summarizing them to coherently fiteither a liberal- or conservative-leaning agenda . The tool suggests hashtag keywords to bolster the polarization of the summary, incase one is inclined to take it to Twitter, Parler or other platforms fortrolling .…

## Topic Scaling A Joint Document Scaling Topic Model Approach To Learn Time Specific Topics

This paper proposes a new methodology to study sequential corpora by implementing a two-stage algorithm that learns time-based topics with respect to a scale of document positions and introduces the concept of Topic Scaling . The first stageranks documents using Wordfish, a Poisson-based document scaling method, toestimate document positions that serve, in the second stage, as a dependent variable to learn relevant topics via a supervised Latent Dirichlet Allocation .…

## Ultra Reliable Indoor Millimeter Wave Communications using Multiple Artificial Intelligence Powered Intelligent Surfaces

A novel framework for guaranteeing ultra-reliable millimeterwave (mmW) communications using multiple artificial intelligence (AI)-enabledreconfigurable intelligent surfaces (RISs) is proposed . The use of multipleAI-powered RISs allows changing the propagation direction of the signalstransmitted from a mmW access point (AP) thereby improving coverage for non-line-of-sight (NLoS) areas .…

## Real Time Global Illumination Using OpenGL And Voxel Cone Tracing

Voxel Cone Tracing, as proposed by Cyril Crassinet al. in 2011, makes use of mipmapped 3D textures containing a voxelizedrepresentation of an environments direct light component to trace diffuse,specular and occlusion cones in linear time to extrapolate a surface fragmentsindirect light emitted towards a given photo-receptor .…

## High Dimensional Differentially Private EM Algorithm Methods and Near Optimal Statistical Guarantees

In this paper, we develop a framework to design differentiallyprivate expectation-maximization (EM) algorithms in high-dimensional latent variable models . We propose a near rate-optimal EM algorithm for low-dimensionallatent variable models in this setting . Simulation studies and real data analysis are conducted to support our results .…

## Enriched Music Representations with Multiple Cross modal Contrastive Learning

Deeplearning is commonly used to obtain representations using various sources of information, such as the audio, interactions between users and songs, or associated genre metadata . In this paper, we present a novel approach that combines multipletypes of information related to music using cross-modal contrastive learning .…

## quantum Case Based Reasoning qCBR

Case-Based Reasoning (CBR) is an artificial intelligence approach toproblem-solving with a good record of success . This article proposes usingQuantum Computing to improve some of the key processes of CBR defining so aQuantum Case-based Reasoning paradigm . The focus is set on designing and implementing a qCBR based on the variational principle that improves itsclassical counterpart in terms of average accuracy, scalability and toleranceto overlapping .…

## GDPR Compliant Blockchains A Systematic Literature Review

Multiple paradoxes between blockchains and GDPR have been highlighted in the recent literature . This article aims to conduct asystematic literature review on GDPR compliant blockchains . The findings synthesized that theblockchains relevant GDPR articles can be categorized into six major groups, including data deletion and modification .…

## Two Truths and a Lie Exploring Soft Moderation of COVID 19 Misinformation with Amazon Alexa

In this paper, we analyzed the perceived accuracy of COVID-19 vaccine Tweets when they were spoken back by a third-party Amazon Alexa skill . We mimicked the soft moderation that Twitter applies to Twitter misinformation content in both forms of warning covers and warning tags to investigate whether the third-partyskill could affect how and when users heed these warnings .…

## Perspective Survey and Trends Public Driving Datasets and Toolsets for Autonomous Driving Virtual Test

Autonomous driving virtual testing has recently gained increasing attention compared with closed-loop testing in real scenarios . The availability and quality of autonomous driving datasets and toolsets are the premise to diagnose theautonomous driving system bottlenecks and improve the system performance .…

## Integrated optimization of heterogeneous network management and the elusive role of macrocells

We consider heterogeneous wireless networks in the physical interferencemodel and introduce a new formulation of the optimization problem underlying their management . This formulation targets the minimization of powerconsumption by integrating base-station activation and many-to-manyassociations into the same mixed-integer nonlinear programming (MINLP) problem .…