Transfer Learning for Human & AI

Microservice architecture refers to the use of numerous small-scale and independently deployed services . It has been a challenge in software engineering to decompose amonolithic system into smaller parts . In this paper, we propose the FeatureTable approach, a structured approach to service decomposition based on the correlation between functional features and microservices .…

Synthetic content analysis is frequently used in nursing research in a wide spectrum of applications . The trend of its use is positive and it is used globally in a variety of research settings . The synthetic content analysis used in our study showed to be a very helpful toolin performing knowledge synthesis, replacing many of the routine activities of conventional synthesis with automated activities this making such studies more economically viable and easier to perform .…

We show open games cover extensive form games with both perfect and imperfect information . Using the former we construct the latter, and subsume previous proposed operators for open games, making progress towards a core, canonical and ergonomic calculus of game operators .…

Cross-Project DefectPrediction (CPDP) has become a popular research topic among researchers . CPDPtrained classifiers based on labeled data sets of one project to predict fault in another project . We propose a Generative Adversarial Network(GAN)-based data transformation to reduce data divergence between source and target projects .…

A neural network trained on reflectance maps of renders of a unit sphere under arbitrary light directionsinfers a low-parameter reflection model representing the reflectance properties of each surface in the scene . Our model is based on a combination ofphenomenological and physics-based scattering models and can relight the scenes from novel viewpoints .…

The recent development of wireless communication has provided many promisingsolutions to emergency response . This article proposes an underwateremergency communication network (UECN) aided by multiple UCLs and autonomousunderwater vehicles to collect underwater emergency data . Simulation results show that the proposed system significantlyimproves the response efficiency and overcomes the limitations of existingworks, which makes contributions to emergency decision-making .…

The field of explainable AI (XAI) has quickly become a thriving and prolific community . A silent, recurrent and acknowledged issue in this area is the lack of consensus regarding its terminology . We propose atheoretical framework that not only provides concrete definitions for theseterms, but it also outlines all steps necessary to produce explanations andinterpretations .…

Decision lists (DLs) find a wide range of uses for classification problems inMachine Learning (ML) This paper shows that computing explanations for DLs is computationallyhard . It proposes propositional encodings for computing abductive explanations (AXps) and contrastive explanations of DLs .…

Open-source model checker Kind 2 can identify Minimal Inductive Validity Cores and Minimal Cut Sets . It can also identify minimal sets of design elements that are sufficient to prove a given set of safety properties . The computed information can be used for several purposes, including assessing the quality of a system specification, tracking the safety impact of model changes .…

In this position paper, we argue for the need to investigate if and how gender stereotypes manifest in search and recommender systems . We focus on how these systems may propagate and reinforce gender stereotypes through their results in learning environments .…

LR-drawing-method is a method of drawing an ordered rooted binary tree . It is based on drawing one root-to-leaf path on a vertical line and attachingrecursively obtained drawings of the subtrees on the left and right . In this paper, we study how to generalize this drawing-method to trees of higher arity.…

Recent graph neural networks integrate node features with network structures, thus enabling inductive node classification models that can be applied to new nodes or even new graphs in the same feature space . Inter-graph differences still exist across graphs within the same domain .…

We show that for every $r$-uniform hypergraph on $m$ edges has a set of atmost $m/2$ edges . We use this to give an algorithm running in time $d^{(1 – \epsilon_r)m}$ that decidessatisfiability of $m-variable $(d, k)$-CSPs can be refuted intree-like resolution in size $2(1)m) The algorithm solves the corresponding #CSP-SAT andMax CSPs .…

Bar-Noy, Motwani and Naor showed that no onlineedge-coloring algorithm can edge color a graph optimally . Their lower bound required bounded-degree graphs, which prompted them to conjecture that better bounds are possible for higher degree graphs . We resolve this thirty-year-old conjecture in the affirmative, presenting a $(1.9+o(1)$-competitive online edge coloring algorithm for general graphs of degree $\Delta = \omega(\log n)$ under vertex arrivals .…

5th generation (5G) mobile networks target the 3.5\,GHz band (3.4 to 3.8GHz) Despite its expected wide usage, there islittle empirical path loss data and mobile radio network planning experience . This paper presents the results of rural,suburban, and urban measurement campaigns using a pre-standard 5G prototypetestbed .…

Learning analytics (LA) is argued to be able to improve learning outcomes,learner support and teaching . However, despite an expanding amount of student (digital) data, there is still little empirical evidence of impact on practice that shows the effectiveness of LA in education settings .…

We conducted a field study at a K-12 private school in the suburbs of Melbourne, Australia . The data capture contained two elements: a 5-monthlongitudinal field study In-Gauge using two outdoor weather stations, as well as indoor weather stations in 17 classrooms .…

Modern railway cars act as Faraday cages and atypical train consist comprises several hundreds of users moving at highvelocity . User expectations have dramatically increased: it isexpected to be online anytime and anywhere . Demand for mobile high-capacity is driven by video and music streaming services, for lower latency and availability by gaming, and for more reliability and even uplinkcapacity by mission critical applications .…

Wireless indoor coverage and data capacity are important aspects of cellularnetworks . With the ever-increasing data traffic, demand for more data capacityindoors is also growing . Newfrequencies foreseen for the 5th generation (5G) of mobile communications inthe millimeter wave (mmWave) spectrum penetrate very poorly into buildings .…

A smart microgrid implemented in Paracas, Ica, Peru, provides electricity to a rural community of 40 families . Real data of solarirradiance, wind speed, energy demand, and voltage of the battery bank from 2periods of operation were studied to find patterns, seasonality, and existingrelations between the analyzed data .…

This paper introduces QAConv, a new question answering (QA) dataset that uses conversations as a knowledge source . We focus on informative conversations including business emails, panel discussions, and work channels . We collect 34,204 QA pairs, including span-based, free-form, and unanswerable questions, from 10,259 selected conversations with both human-written and machine-generated questions .…

Modern rock music proved to bedifficult to estimate tempo and chord recognition did not work . Using a neural network has been one of the simplest ways of dealing with it . The recognition process is time-consuming centred centred on onextremely complicated and memory-intensive methods .…

An effective method for cross-lingual transfer is to fine-tune a bilingual ormultilingual model on a supervised dataset in one language and evaluate it on another language in a zero-shot manner . Translating examples at training time or inference time are also viable alternatives .…

Classical simulations of quantum computations are vital for the futuredevelopment of this emerging technology . Decision diagrams have been proposed as a complementary technique which frequently allows to tackle the inherent exponential complexity of these simulations . In this work, we show that both problems can be tackled together by employing a hybrid Schrodinger-Feynman scheme for the simulation.…

In this paper we present a quantum algorithm which increases the amplitude of the states corresponding to the solutions of the search problem by a factor of almost two .…

We show that M-subshift/SFT systems are precisely the expansive dynamical dynamical M-systems . We also study theautomorphism groups and periodic points of sofically presented systems . A keyidea is the introduction of so-called automatic spaces . We show any finite connected simplicial complex is aconnected component of a finitely presented system, and prove that conjugacy of one-dimensional dynamical systems is undecidable.…

A neural radiance field (NeRF) is a scene model supporting high-quality viewsynthesis, optimized per scene . In this paper, we explore enabling user editing of a category-level NeRF . We introduce a method for propagating coarse2D user scribbles to the 3D space, to modify the color or shape of a local region .…

The lack of specifications is a key difference between software engineering and machine learning . We discuss how it drastically impacts how wethink about divide-and-conquer approaches to system design . We faceweak specifications, wrong specifications, and unanticipated interactions among components and their specifications .…

Relation algebra and its reducts provide us with a strong tool for reasoning about nondeterministic programs and their partial correctness . We formalise the framework for relational reasoning about total correctness using semigroups with ordinary composition and demonicic lattice operations .…

Robot Operating System (ROS) is a popular robotics middleware framework . It underwent a redesign and reimplementation under the nameROS~2 . It now features QoS-configurable communication and a flexible layeredarchitecture . Micro-ROS is a variant developed specifically for resource-constrained microcontrollers (MCU) Such MCUs are commonly used inrobotics for sensors and actuators, for time-critical control functions, and for safety .…

Closure spaces are a generalisation of topological spaces obtained by removing idempotence requirement on the closure operator . We adapt the notion of bisimilarity for topological models, namelyTopo-bisimilarity, to closure models . We also address the issue of (space) minimisation via the three equivalences .…

Coordination is often critical to forming prosocial behaviors . State of the art reinforcement learning algorithmsoften suffer from converging to socially less desirable equilibria . We propose using a less restrictive peer-rewarding mechanism,gifting . Gifting allows each agent to give some of their reward to other agents .…

This paper introduces reproducible research, and explains its importance,benefits and challenges . Some important tools for conducting reproducibleresearch in Transportation Research are also introduced . The sourcecode for generating this paper has been designed in a way so that it can be used as a template for researchers to write their future journal papers .…

Reservoir computing (RC) offers efficient temporal data processing with a low training cost by separating recurrent neural networks into a fixed network with recurrent connections and a trainable linear network . The quality of the fixed network, called reservoir, is the most important factor that determines the performance of the RC system .…

In this paper we discuss the risk model of the German Corona Warning App(CWA) in two versions . Both are based on a general semi-quantitative risk approach that is not state of the art anymore . However, it turns out that the CWA uses a much more limited model, that does not even assess risk, but relies only on one parameter, aweighted exposure time .…

DoS and DDoS attacks have been growing in size and number over the last decade . Existing solutions to mitigate these attacks are in generalinefficient . Two methods based on the ability of VariationalAutoencoders to learn latent representations from network traffic flows are proposed .…

In the (fully) dynamic set cover problem, we have a collection of $m$ sets from a universe of size $n$ that undergo element insertions and deletions . The goal is to maintain an approximate set cover of the universe after each update .…

Wireless body area networks (WBANs) are becoming increasingly popular as they allow individuals to monitor their vitals remotely from the hospital . With the spread of the SARS-CoV-2pandemic, the availability of portable pulse-oximeters and wearable heart ratedetectors has boomed in the market .…

This chapter reflects on the history and trajectory of urban analytics as a scholarly and professional discipline . It argues that privacy and ethical concerns are toooften ignored as ubiquitous monitoring and analytics can empower socialrepression . It concludes with a call for a more critical urban analytics that emphasizes human dignity and learns from and supports marginalized communities in the field of data analysis .…

Traditionally, heuristic methods are used to generate candidates for largescale recommender systems . Model-based candidate generation promises multiplepotential advantages, primarily that we can explicitly optimize the same objective as the downstream ranking model . We first explore theynamics of the dataset bias problem and then demonstrate how to use randomsampling techniques to mitigate it .…

The undecidability of basic decision problems for general FIFO machines such as reachability and unboundedness is well-known . In this paper, we provide anunderapproximation for the general model by considering only runs that are input-bounded . We prove, by reducing this model to acounter machine with restricted zero tests, that the rational-reachabilityproblem is decidable .…

Equity of educational outcome and fairness of AI with respect to race have been topics of increasing importance in education . We find that an adversarial learning approach, combined with grade label balancing,achieved by far the fairest results . With AI-infused technology supportsincreasingly prevalent on campuses, our methodologies fill a need forframeworks to consider performance trade-offs withrespect to sensitive studentattributes and allow institutions to instrument their AI resources in ways that are attentive to equity and fairness .…

The window mechanism was introduced by Chatterjee et al. to reinforce mean-payoff and total-payoffs objectives with time bounds in two-playerturn-based games on graphs . We show that checking that all time-divergent paths of atimed automaton satisfy such a window parity objective can be done inpolynomial space, and that the corresponding timed games can be solved inexponential time .…

We consider the problem of estimating quantum observables on a collection of qubits . We introduce estimators based on randomised measurements . They use decision diagrams to sample from probability distributions on measurement bases . This approach generalises previously knownuniform and locally-biased randomised estimators .…

Study of multi-type Protein-Protein Interaction (PPI) is fundamental for understanding biological processes from a systematic perspective and revealing disease mechanisms . Existing methods suffer from significant performancedegradation when tested in unseen dataset . We propose a graph neural network based method for better inter-novel-protein interaction prediction .…

Cycle representatives of persistent homology classes can be used to provide descriptions of topological features in data . The non-uniqueness of these representatives creates ambiguity and can lead to many differentinterpretations of the same set of classes . We conduct theseoptimizations via standard linear programming methods, applying general-purposesolvers to optimize over column bases of simplicial boundary matrices .…

Community search aims at finding densely connected subgraphs for queryvertices in a graph . We theoretically prove this problem is NP-hard and analyze its non-approximability . To efficiently tackle the problem, we develop a heuristic algorithm, which first finds a BCCcontaining the query vertices, then removes the farthest verticesto those vertices from the graph .…

We present a partitioned neural network-based framework for learning offluid-structure interaction (FSI) problems . We decompose the simulation domain into two smaller sub-domains, i.e., fluid and solid domains, and incorporate an independent neural network for each . Aquasi-Newton method is used to accelerate the FSI coupling convergence .…

We present a unified and hardware efficient architecture for two stage voicetrigger detection (VTD) and false trigger mitigation (FTM) tasks . We propose a streaming transformer (TF) encoder architecture, which progressively processes incoming audio chunks and maintains audio context . The proposed model yields an average 18% relative reduction in false reject rate (FRR) for the VTD task at a given false alarm rate .…

This paper is part of an ongoing endeavor to bring the theory of fairdivision closer to practice by handling requirements from real-life applications . We prove upper and lower boundson achievable maximin share guarantees when the usable shapes are squares, fatrectangles, or arbitrary axes-aligned rectangles .…