Transfer Learning for Human & AI

Test-driven development (TDD) is an agile software developmentapproach that has been widely claimed to improve software quality . The extent to which TDD improves quality appears to be largely dependent upon the characteristics of the study in which it is evaluated .…

The mainingredient of the construction is a subconstruction whose points are in convexposition . We prove that for every $m$ there is a finite point set $\mathcal{P}$ in theplane such that no matter how $p$ is three-colored, there is always a disk containing exactly $m $points, all of the same color .…

A multi-agent simulation tool simulates the behavior of skiers in a ski-area . This is the first tool able to simulate and compare management and planning scenarios as well as their impact on skier waiting times . An addition of 1620 new skiers delays the skier average waiting time by 12 pourcents .…

Relation extraction (RE) consists in identifying and structuringautomatically relations of interest from texts . Among tested architectures and strategies, our*BERT-segMCNN with finetuning reaches performances higher than the state-of-the-art on the two corpora (1.73 % and 32.77 % absolute improvement onChemProt and PGxCorpus corpora respectively) More generally, our experimentsillustrate the expected interest of fine-tuning with BERT, but also theunexplored advantage of using structural information (with sentencesegmentation), in addition to the context classically leveraged by BERT.…

Sentence Encoder Assembly (SEA) supports text-video matching in multiple encoder-specific commonspaces . SEA surpasses the state-of-the-art. In addition, SEA is extremely ease to implement. SEA is an appealing solution for AVS and promising forcontinuously advancing the task by harvesting new sentence encoders.…

Machine learning has the potential to aid in mitigating the human effects of climate change . Previous applications of machine learning to tackle the humaneffects in climate change include informing individuals of their carbon footprint and strategies to reduce it .…

The acceleration race of digital computing technologies seems to be steeringtoward impasses — technological, economical and environmental — a condition that has spurred research efforts in alternative, “neuromorphic” (brain-like)computing technologies . The idea of exploiting nonlinear physical phenomena “directly” for non-digital computing has been explored under names like “unconventional computing”, “natural computing”,”physical computing”, or “in-materio computing” I stake out how a general concept of “computing” can bedeveloped which comprises digital, neuromorphic, unconventional and possible future paradigms .…

The newly emerged machine learning (e.g. deep learning) methods have become a driving force to revolutionize a wide range of industries, such as smarthealthcare, financial technology, and surveillance systems . Privacy has emerged as a big concern in this machine learning-based artificialintelligence era .…

Named Entity Recognition (NER) is a fundamental task in the fields of naturallanguage processing and information extraction . NER has been widely used as a tool or an essential component in a variety of applications such as question answering, dialogue assistants and knowledge graphs development .…

The aim is to upscale theelasto-plastic behavior for a representative volume of the material . The validity of sucheigen-strain based formulation of multiscale elastic media is hereinexamined by comparing its predictions against Finite Element (FE) simulations .Considering the plastic strains as eigen-stressed elastic media allows us to employ the tools offered by Continuum Micromechanics which are mainly developed .…

The differences in the content of the magazines aimed specifically at women or men are a means to create and reproduce gender stereotypes . Cultural products are a source to acquire individual values and behaviours . The frequency of the topics Family, Business and Women as sex objects, present an initial bias that tends to disappear over time .…

The closed-loop performance of model predictive controllers (MPCs) issensitive to the choice of prediction models, controller formulation, and tuning parameters . MPC tuning is typically donemanually to satisfy (probabilistic) constraints . In this work, we demonstrate ageneral approach for automating the tuning of MPC under uncertainty .…

InTAS roadtopology accurately represents Ingolstadt real road map . Twenty trafficlights systems were simulated according to the real program deployed on the traffic lights . Traffic demand was modeled based on the activitygen method,considering demographic data and real-traffic information . The city public transport system was also simulated accordingly to bus time-tables and their routes .…

We propose a new model that describes the dynamics of epidemic spreading on connected graphs . Our model consists in a PDE-ODE system where at each vertices we have a standard SIR model and connexions between vertices are given by heat equations on the edges .…

This paper investigates the communication strategy for second-ordermulti-agent systems with nonlinear dynamics . To save the scarce resources of communication channels, a novel event-triggered communication mechanism is designed without using continuous signals among the followers . The results are verified through numericalexamples .…

Query-by-example (QbE) speech search is the task of matching spoken utterances within a search collection . In low- or zero-resource settings, QbE search is often addressed with approaches based on dynamic time warping . Recent work has found that methods based on acoustic word embeddings(AWEs) can improve both performance and search speed .…

Contextualized representations of language are typically motivated by their ability to encode context . We introduce an approach to address this question using Representational Similarity Analysis (RSA) In all cases, we show that BERT’s contextualized embeddings reflect thelinguistic dependency being studied .…

A fundamental challenge in learning to control an unknown dynamical system is to reduce model uncertainty by making measurements while maintaining safety . In our framework, the state of the system is required to stay within a given safety region under the (possibly repeated) action of all systems that are consistent with the information gathered so far .…

The solution for this problem can besolved using a decentralized database to hold the hashes and the firmware . This paper discusses the possible implications of insecure servers used to host thefirmwares of commercial IoT products . It aims to provide a blockchain baseddecentralized solution to host firmware files with the property ofimmutability, and controlled access to the firmware upload functions so as to stop unauthorized use .…

A data-fusion-assisted telemetrylayer between the physical layer and control layer is proposed in this paper . The data fusion methodologies are elaborated on three different levels: SourceLevel, Space Level and Model Level . For each level, various data fusionalgorithms are introduced and relevant works are reviewed .…

Contract scheduling is a general technique that allows to design a system with interruptible capabilities, given an algorithm that is not necessarilyinterruptible . Previous work on this topic has largely assumed that the interruption is a worst-case deadline that is unknown to the scheduler .…

In this study we analyse the key driving factors of preprints in enhancingscholarly communication . Articles with preprint versions are morelikely to be mentioned in social media and have shorter Altmetric attentiondelay . We could observe that the “early-view” and “open-access” effects of preprint” of .…

Neuronal circuits internally regulate electrical signaling via a host of homeostatic mechanisms . Two prominent mechanisms, synaptic scaling andstructural plasticity, are believed to maintain average activity within an operating range . However, both mechanisms operate on relatively slow timescales and thus face fundamental limits due to delays .…

The verification of concurrent programs remains an open challenge due to thenon-determinism in inter-process communication . The reads-from (RF) equivalence was recently shownto be coarser than the Mazurkiewicz equivalence, leading to impressivescalability improvements for SMC under SC . For TSO and PSO, the standard equivalence has been Shasha-Snir traces .…

We consider the problem of identifying discontinuous doping profiles insemiconductor devices from data obtained by different models connected to thevoltage-current map . Stationary as well as transient settings are discussed . Numericalimplementations for the so-called stationary unipolar and stationary bipolarcases show the effectiveness of a level set approach to tackle the inverse problem .…

Stochastic sparse adversarial attacks (SSAA) are simple, fast and purely noise-based targeted and untargeted attacks of NNC . SSAA offer new examples of sparse (or $L_0$) attacks for which only few methodshave been proposed previously . These attacks are devised by exploiting asmall-time expansion idea widely used for Markov processes .…

Energy-Based Models (EBMs) have a natural way to support a dynamically-growing number of tasks or classes that causes less interference with previously learned information . EBMs outperform the baseline methods by a large margin on several continual learningbenchmarks . We also show that EBMs are adaptable to a more general continuallearning setting where the data distribution changes without the notion of explicitly delineated tasks .…

When using heterogeneous hardware, barriers of technical skills such as OpenMP, CUDA and OpenCL are high . I have proposedenvironment-adaptive software that enables automatic conversion, configuration . However, there has been no research toproperly and automatically offload the mixed offloading destination environmentsuch as GPU, FPGA and many core CPU.…

Particle swarm optimization (PSO) method cannot be directly used in the problem of hyper-parameter estimation . Bayesian optimization (BO) framework is capable of converting the optimization of an acquisition function . The proposed method in this paper uses theparticle swarm method to optimize the acquisition function in the BO framework .…

A SpatialPoint Transformer is first introduced to map the input local surface into acarefully designed cylindrical space, enabling end-to-end optimization withSO(2) equivariant representation . A Neural Feature Extractor leverages the powerful point-based and 3D cylindrict convolutional neural layers to derive a compact and representative descriptor for matching .…

This paper proposes a novel multilevel inverter for automotive applications . The topology consists of a modular DCDC converter and a tap selector . The DC-DC converter is capable of self-balancing its modules and thus does not require large capacitors which yields a high power density .…

The General Purpose Atomic Crosschain Transaction protocol allows composableprogramming across multiple Ethereum blockchains . It allows for inter-contractand inter-blockchain function calls that are both synchronous and atomic . If one part fails, the whole call graph of function calls is rolled back .…

We analyze and experimentally compare various rotational error metrics for use in quadrotor controllers . We provide a catalog of proposed rotational metrics, place them into the same framework . We show experimental results to highlight the salient differences between the rotational errors .…

A cryptographic protocol (CP) is a distributed algorithm designed to provide a secure communication in an insecure environment . Errors in the CPs can lead to great financial and social damage, therefore it is necessary to use mathematical methods to justify the correctness and safety of the CP .…

In this article, we develop and analyze a finite element method with the first family N\’ed\’elec elements of the lowest degree for solving a Maxwellinterface problem . We establish a few important properties for the IFEfunctions including the unisolvence according to the edge degrees of freedom, the exact sequence relating to the $H^1$ IFE functions and the optimalapproximation capabilities .…

Imagery is frequently used to model, represent and communicate knowledge . Causality is defined in terms of precedence (thecause precedes the effect), concurrency (often, an effect is provokedsimultaneously by two or more causes) and circularity (a cause provokes the effectand the effect reinforces the cause) We will introduce Fuzzy Stochastic Timed Petri Nets as a graphical tool able to represent time, co-occurrence, loopingand imprecision in causal flow .…

A cascaded speech translation model relies on discrete and non-differentiabletranscription . Such modelingsuffers from error propagation between ASR and MT models . Our experiments on four tasks with different datascenarios show that the model outperforms cascade models up to 1.8% in BLEU and 2.0% in TER .…

Multi-task benchmarks such as GLUE and SuperGLUE have driven great progressof pretraining and transfer learning in Natural Language Processing (NLP) These benchmarks mostly focus on a range of Natural Language Understanding(NLU) tasks, without considering the Natural Language Generation (NLG) models .…

Paper explores unconventional method-generating numerical and non-numerical asset condition data based on condition degradation, condition correlation and categorical distribution models . Empirical knowledge from human experts can also be incorporated in themodeling process . Method can be used to conveniently generatehypothetical data for research purposes .…

PeleNetframework aims to simplify reservoir computing for neuromorphic hardware . It is build on top of the NxSDK from Intel and is written in Python . The framework manages weight matrices, parameters and probes . With this, the user is not confronted with technical details and can concentrate on experiments .…

This letter investigates parallelism approaches for equations and Jacobianevaluation in power flow calculations . Case studies on the 70,000-bus synthetic grid show that equation evaluations can be accelerated by ten times, and the overall Newton power flow outperforms MATPOWER by 20% .…

Inspired by InstaHide challenge [Huang, Song, Li and Arora’20], [Chen, Songand Zhuo’20] recently provides one mathematical formulation of InstaHash attackproblem under Gaussian images distribution . They show that it suffices to use $O(n_{\mathsf{priv) +\mathrm{poly}(n__priv)$ samplesto recover one private image in $n_n_priv_time for any integer $k_\maths$ – 2/(k_maths) + 1)$ .…

For safety-critical autonomous driving tasks, it is desirable to monitor in operation time if the input for the DNN is similar to the data used in DNN training . The algorithm performs a soundworst-case estimate of neuron values with inputs (or features) subject toperturbation, before the abstraction function is applied to build the monitor .…

Serverless architecture is more frequently associated with the architectural style for developing cloud-native applications . Blockchains are distributed systems designed to enable collaborative scenarios involving untrusted parties . The decentralizedpeer-to-peer nature of blockchains makes it interesting to consider them inserverless architectures, since resource allocation and management tasks are not required to be performed by users .…

We present algorithms that extend the path-based hierarchical drawing framework and give experimental results . Our algorithms run in $O(km)$ time,where $k$ is the number of paths and $m$ is number of edges of the graph . We also provide somecomparison to a well known .…

Android Apps are frequently updated, every couple of weeks, to keep up withchanging user, hardware and business demands . Correctness of App updates ischecked through extensive testing . Recent research has proposed tools forautomated GUI event generation in Android Apps .…

A new primal-dual weak Galerkin (PDWG) finite element method is introduced . The PDWG scheme yields a system of equations involving both theprimal equation and the adjoint (dual) equation . The optimal order errorestimate for the primal variable in a low regularity assumption is established .…

The Winograd Schema Challenge (WSC) is a common-sense reasoning task that requires background knowledge . In this paper, we contribute to tackling WSC infour ways . We suggest a keyword method to define a restricted domain where distinctive high-level semantic patterns can be found .…

The vast majority of mesh-based modelling applications iteratively transform mesh vertices under prescribed geometric conditions . In current PBD implementations, the area preservation constraint is linearised . The solution comes out through the iterations, without a guarantee of optimality, and the process may fail for degenerate inputs where thevertices are colinear or colocated .…

This paper presents a new deep learning-based model for fault detection in offshore wind turbines . The proposed model uses the nonlinear relationships among multiple sensor variables and thetemporal dependency of each sensor on others that considerably increases the performance of the model .…