Memory management is still important especially in mobiledevices with small memory capacity . Low memory killer (LMK) and out-of-memory killer (OOMK) are widely used memory management solutions in mobile systems . They forciblyterminate applications when the available physical memory becomes insufficient .…
Evaluating Multilingual Text Encoders for Unsupervised Cross Lingual Retrieval
Pretrained multilingual text encoders based on neural Transformerarchitectures, such as multilingual BERT (mBERT) and XLM, have achieved strong performance on a myriad of language understanding tasks . However, questions remain to which extent this generalizes to unsupervised settings and for ad-hoc cross-lingualIR (CLIR) tasks .…
Joint Autoregressive and Graph Models for Software and Developer Social Networks
Social network research has focused on hyperlink graphs, bibliographiccitations, friend/follow patterns, influence spread, etc. Large softwarerepositories form a highly valuable networked artifact, usually in the form of a collection of packages, their developers, dependencies among them,and bug reports . This “social network of code” is rarely studied by social network researchers .…
Rethink Training of BERT Rerankers in Multi Stage Retrieval Pipeline
Rerankers fine-tuned from deep LM estimates candidate relevance based on rich contextualized matching signals . Meanwhile, deep LMs can also beleveraged to improve search index, building retrievers with better recall . We propose a Localized Contrastive Estimation (LCE) for trainingrerankers and demonstrate it significantly improves deep two-stage models .…
Multimodality in VR A survey
Virtual reality has the potential to change the way we create and consumecontent in our everyday life . Entertainment, training, design andmanufacturing, communication, or advertising are all applications that already benefit from this new medium reaching consumer level . In VR, like inthe real world, users integrate the multimodal sensory information they receive to create a unified perception of the virtual world .…
Personalised Recommendations in Mental Health Apps The Impact of Autonomy and Data Sharing
The recent growth of digital interventions for mental well-being prompts acall-to-arms to explore delivery of personalised recommendations from a user’s perspective . In a randomised placebo study with a two-way factorial design, we analysed the difference between an autonomous user experience as opposed to personalised guidance .…
UNIT Unifying Tensorized Instruction Compilation
Tensorized instructions formixed-precision tensor operations, like Intel VNNI, Tensor Core, and ARM-DOT . These instructions involve a computing idiom that reduces multiple lowprecision elements into one high precision element . The generated end-to-end inference model achieves 1.3x speedup over Intel oneDNN on an x86 CPU and 1.13x speed up over TVM solution for ARM DOT on an ARM CPU .…
Mechanism Design for Cumulative Prospect Theoretic Agents A General Framework and the Revelation Principle
This paper initiates a discussion of mechanism design when the participatingagents exhibit preferences that deviate from expected utility theory . In particular, we show that the revelation principle, which hastraditionally played a fundamental role in mechanism design, does not continue to hold under CPT .…
A Joint Diagonalization Based Efficient Approach to Underdetermined Blind Audio Source Separation Using the Multichannel Wiener Filter
This paper presents a computationally efficient approach to blind sourceseparation (BSS) of audio signals . When there are as manysources as microphones as microphones (i.e., the determined case), BSS can be performedcomputationally efficiently by independent component analysis (ICA) However, ICA is basically inapplicable to the underdeterminedcase .…
Assessing the Benefits of Model Ensembles in Neural Re Ranking for Passage Retrieval
Modelensembling can indeed benefit the ranking quality, particularly with supervisedlearning-to-rank although also with unsupervised rank aggregation . Tests with the MS-MARCO dataset show that modelensembled can benefit the rankings quality . We use a previous technique named FastGeometric Ensembling to generate multiple model instances from particulartraining schedules, then focusing or attention on different types of approaches to combining the results from the multiple models .…
Model based Policy Search for Partially Measurable Systems
Monte Carlo Probabilistic Inferencefor Learning COntrol for Partially Measurable Systems (MC-PILCO4PMS) relies onGaussian Processes (GPs) to model the system dynamics, and on a Monte Carlo approach to update the policy parameters . The effectiveness of the proposed algorithm has been tested both in simulation and in two real systems .…
A two stage data association approach for 3D Multi object Tracking
Multi-object tracking (MOT) is an integral part of any autonomous drivingpipelines because it produces trajectories which has been taken by other moving objects in the scene and helps predict their future motion . Track-by-detection has become the dominant paradigm in 3D MOT .…
Machine Learning Based Early Fire Detection System using a Low Cost Drone
This paper proposes a new machine learning based system for forest fireearlier detection in a low-cost and accurate manner . The common problem in the prevalent algorithms used in firedetection is the high false alarm and overlook rates. Confirming the resultobtained from the visualization with an additional supervision stage will increase the reliability of the system as well as guarantee the accuracy of the result.…
E commerce warehousing learning a storage policy
E-commerce with major online retailers is changing the way people consume . The goal of increasing delivery speed while remaining cost-effective posessificant new challenges for supply chains as they race to satisfy thegrowing and fast-changing demand . We propose formulating this system as a Partially Observable Markov Decision Process, and using a Deep Q-learningagent from Reinforcement Learning, to learn an efficient real-time storagepolicy that leverages repeated experiences and insightful forecasts .…
The Diagnosis of Asthma using Hilbert Huang Transform and Deep Learning on Lung Sounds
Lung auscultation is the most effective and indispensable method for diagnosing various respiratory disorders by using the sounds from the airways during inspirium and exhalation using a stethoscope . The proposed DBN separated lung sounds from asthmatic and healthy subjects with high classification performance rates of 84.61%, 85.83%, and 77.11% for overall accuracy, sensitivity, andselectivity, respectively using frequencytime analysis .…
Complete trace models of state and control
We consider a hierarchy of four typed call-by-value languages with either higher-order references or callcc or no control operator . The paper provides a systematic development ofoperational game semantics for all four cases, which represent the state-based face of the so-called semantic cube .…
Synwalk Community Detection via Random Walk Modelling
Synwalk builds upon a solid theoretical basis and detectscommunities by synthesizing the random walk induced by the given network from a class of candidate random walks . We thoroughly validate the effectiveness ofour approach on synthetic and empirical networks, respectively, and compare Synwalk’s performance with the performance of Infomap and Walktrap .…
Content Based Textual File Type Detection at Scale
Programming language detection is a common need in the analysis of largesource code bases . We propose asimple model that (a) use a language-agnostic word tokenizer for textual files, (b) group tokens in 1-/2-grams, (c) build feature vectors based on N-gramfrequencies .…
Privacy Preserving Distributed Optimal Power Flow with Partially Homomorphic Encryption
We propose a privacy-preserving distributed optimal powerflow (OPF) algorithm based on partially homomorphic encryption (PHE) We exploit the alternating direction method of multipliers (ADMM) to solve the OPF in a distributed fashion . The SDP can be solved locally with only the sign messages from neighboring agents, which preservest the privacy of the primal update .…
Designing a Reliable Inland Waterway Transportation Network under Uncertainty
Study aims at developing areliable inland waterway transportation network considering interactions between different transportation entities . Inland waterway transport network significantly supports the overallfreight transportation of the nation . Acapacitated multi-commodity, multi-period, stochastic, two-stage mixed-integerlinear programming (MILP) model is proposed .…
Online End to End Neural Diarization Handling Overlapping Speech and Flexible Numbers of Speakers
The end-to-end neuralspeaker diarization (EEND) model has already achieved significant improvement compared with conventional clustering-based methods . However, the original EEND has two limitations: i. EEND does not perform well in online scenarios;ii.ii) the number of speakers must be fixed in advance .…
Monitoring nonstationary processes based on recursive cointegration analysis and elastic weight consolidation
Traditional approaches generally misidentify the normal dynamic deviations as faults and thus lead to high false alarms . In this paper, recursive cointegrationanalysis (RCA) is first proposed to distinguish the real faults from normalsystems changes . When the systementers a new operating condition, the RCA-RPCA model is rebuilt to deal with the new condition .…
Factors that Affect Software Systems Development Project Outcomes A Survey of Research
Determining the factors that have an influence on software systemsdevelopment and deployment project outcomes has been the focus of extensive andongoing research for more than 30 years . We provide here a survey of theresearch literature that has addressed this topic in the period 1996-2006 .…
Work Patterns of Software Engineers in the Forced Working From Home Mode
The COVID-19 outbreak has caused a major disruption worldwide . But what happened to companies developing digital services? Were they interrupted as much or at all? And how has the enforcedWorking-From-Home (WFH) mode impacted their ability to deliver software? We hear that some managers are concerned that their engineers are not working effectively from home, or even lack the motivation to work in general, that teams lose touch and that managers do not notice when things go wrong .…
Compositional Construction of Abstractions for Infinite Networks of Discrete Time Switched Systems
In this paper, we develop a compositional scheme for the construction ofcontinuous approximations for interconnections of infinitely many discrete-timeswitched systems . An approximation (also known as abstraction) is itself acontinuous-space system, which can be used as a replacement of the original system in a controller design process .…
Effect of Window Size for Detection of Abnormalities in Respiratory Sounds
The duration of the sounds used in the diagnosis affects the speed of the diagnosis . In this study, theeffect of window size on diagnosis of abnormalities in respiratory sounds and the most efficient recording time for diagnosis were analyzed .…
Commutative Event Sourcing vs Triple Graph Grammars
This paper proposes Commutative Event Sourcing as a simple and reliablemechanism for model synchronisation, bidirectional model to modeltransformations, incremental updates, and collaborative editing . It is a restricted form of a Triple Graph Grammar where the rules are either overwriting or commutative .…
Walking Through the Method Zoo Does Higher Education really meet Software Industry Demands
This paper analyzes the current state of practice in software engineering education . We want to compare contemporary education with industrial practice to understand if frameworks, methods and practices for software and system development taught at HEIs reflect industrial practice .…
Learning rich touch representations through cross modal self supervision
The sense of touch is fundamental in several manipulation tasks, but rarely used in robot manipulation . In this work we tackle the problem of learning richtouch features from cross-modal self-supervision . We evaluate them identifying objects and their properties in a few-shot classification setting .…
A computer algebra system for the study of commutativity up to coherent homotopies
ComCH is a lightweight specialized computer algebra system . It provides models for well known objects, the surjection and Barratt-Ecclesoperads . ComCH provides effective constructions of Steenrod cohomology operations at all prime . The primary examples treated by ComCH are the cochain complexes of spaces .…
Streaming from the Air Enabling High Data rate 5G Cellular Links for Drone Streaming Applications
Enabling high data-rate uplink connectivity for drones is achallenging problem . A flying drone has a higher likelihood of having line-of-sight propagation to base stations . This may result in uplink inter-cell interference and performance degradation for the neighboring ground UEs .…
Turkish Voice Commands based Chess Game using Gammatone Cepstral Coefficients
Voice recordings were taken from 50 people (23 men and 27 women) While recording the sound, 29 words from each person were used which are determined as necessary for playing the game . Two different classification procedures were applied, namely, person-based and general .…
On the Use of Computational Fluid Dynamics CFD Modelling to Design Improved Dry Powder Inhalers
Computational Fluid Dynamics (CFD) simulations are performed to investigate the impact of adding a grid to a two-inlet dry powder inhaler . The purpose of the paper is to show the importance of the correct choiceof closure model and modeling approach, as well as to perform validation against particle dispersion data obtained from in-vitro studies and flowvelocity data .…
Multi robot energy autonomy with wind and constrained resources
One aspect of the ever-growing need for long term autonomy of multi-robotsystems is ensuring energy sufficiency . In scenarios where charging facilities are limited, battery-powered robots need to coordinate to share access . In this work we extend previous results by considering robotsthat carry out a generic mission while sharing a single charging station, while being affected by air drag and wind fields .…
Deductive Verification of Floating Point Java Programs in KeY
Deductive verification has been successful in verifying interesting properties of real-world programs . One notable gap is the limited support for floating-point reasoning . Floating-point arithmetic is unintuitive to reason about due to rounding and the presence of special values infinity and NaN .…
Qualitative Research on Software Development A Longitudinal Case Study Methodology
This paper reports the use of a qualitative methodology for conducting longitudinal case study research on software development . Our aim is to illustrate the utility of such research as a complement to existing methodologies for studying software development, so as to enable the community to develop afuller and richer understanding of this complex, multi-dimensional phenomenon .…
Adversarial Machine Learning for Flooding Attacks on 5G Radio Access Network Slicing
Network slicing manages network resources as virtual resource blocks for the 5G Radio Access Network (RAN) Each communication request comes with quality of experience (QoE) requirements such as throughput and latency . The reward is maximized over time by allocating resources, e.g.,…
Boost then Convolve Gradient Boosting Meets Graph Neural Networks
Graph neural networks (GNNs) are powerful models that have been successful invarious graph representation learning tasks . Previous GNN models have mostly focused on networks withhomogeneous sparse features . In this work, we propose a novel architecture that trains GBDT and GNN jointly to get the best of both worlds: the GBDT modeeals with heterogeneous features, while GNN accounts for the graph structure .…
Ten Simple Rules for Attending Your First Conference
Conferences are a mainstay of most scientific disciplines, where scientists come together to share cutting-edge ideas and approaches . The authors of this piece have attended their fair share of conferences and have mentored hundreds of students in understanding the “unwrittenrules” and pro-tips of conference attendance .…
A Gauss Seidel projection method with the minimal number of updates for stray field in micromagnetic simulations
Magnetization dynamics in magnetic materials is often modeled by theLandau-Lifshitz equation . Inmicromagnetic simulations, the computational cost relies heavily on thetime-marching scheme and evaluation of stray field . Explicit marchingschemes are efficient but suffer from severe stability constraints . GSPM-BDF2 updates the stray field only once, leading to an efficiency improvement of about $60\%$ than the state-of-the-art GSPm for micromagnetic simulation .…
Effect of Deep Learning Feature Inference Techniques on Respiratory Sounds
Analysis of respiratory sounds increases its importance every day . New techniques are continuing to be developed to improve these methods . Image filters were applied to thevalues obtained from audio signals and the results of the features formed from this were examined in machine learning and deep learning techniques .…
BOOSTR A Dataset for Accelerator Control Systems
BOOSTR (Booster Operation Optimization Sequential Time-Series for Reinforcement) was created to provide cycle-by-cycle time series of readings and settings from instruments and controllable devices of the Booster, Fermilab’s Rapid-Cycling Synchrotron (RCS) operating at 15 Hz . To our knowledge, this is the first known dataset of accelerator device device parameters made publicly available .…
A Critical Comparison on Six Static Analysis Tools Detection Agreement and Precision
Developers use Automated Static Analysis Tools (ASATs) to controlfor potential quality issues in source code . We analyze 47 Java projects and derive a taxonomy of warnings raised by 6 state-of-the-practice ASATs . To assess their agreement, we compared them by manually analyzing – at line-level – whether they identify the same issues .…
A Fragile multi CPR Game
A Fragile CPR Game is an instance of a resource sharing game where a common-pool resource is shared among multiple players . Each player has a fixed initial endowment and is faced with the task of investing in the resource without forcing it to fail .…
User Aware Power Management for Mobile Devices
The power management techniques to extend battery lifespan is becomingincreasingly important due to longer user applications’ running time in mobiledevices . Even when users do not use any applications, battery lifespandecreases continually . In this paper, we propose anew power management system that recognizes idle time of the device to reduce the battery consumption of mobile devices .…
Physical Reservoir Computing with Origami and its Application to Robotic Crawling
A new paradigm called physical reservoir computing has recently emerged, where the nonlinear dynamics of high-dimensional and fixed physical systems areharnessed as a computational resource to achieve complex tasks . This study uncovers thelinkages between the origami reservoir’s physical designs and its computingpower, offering a guideline to optimize the computing performance .…
Comparison and Improvement for Delay Analysis Approaches Theoretical Models and Experimental Tests
Computer network tends to be subjected to the proliferation of mobile demands and increasingly multifarious . In this paper, we describe an analytical model based on the measurement for the delay of eachpacket passing through the single existing routers in the network environment .…
Centralized Collision free Polynomial Trajectories and Goal Assignment for Aerial Swarms
Computationally tractable methods are developed for centralized goalassignment and planning of collision-free polynomial-in-time trajectories for multiple aerial robots . The method first assigns robots to goals to minimize total time- in-motion based on initial trajectories . The plans are then refined by checking for potential collisions and resolving them using either start time delays or altitude assignment .…
Location Management in IP based Future LEO Satellite Networks A Review
Future integrated terrestrial, aerial, and space networks will involvethousands of Low Earth Orbit (LEO) satellites forming a network ofmega-constellations . We present acomprehensive and critical review of the state-of-the-art research in LEOSatNets location management . We give an overview of the InternetEngineering Task Force (IETF) mobility management standards (e.g.,…
Arabic Speech Recognition by End to End Modular Systems and Human
Recent advances in automatic speech recognition (ASR) have achieved accuracy levels comparable to human transcribers . Previous work focused on the Englishlanguage and modular hidden Markov model-deep neural network (HMM-DNN) systems . For ASR the end-to-end work led to 12.5%, 27.5% WER; a new milestone for the MGB2, MGB3, and MGB5 challenges respectively .…