Showing papers by "French Institute for Research in Computer Science and Automation published in 2021"

Advances and open problems in federated learning

Abstract: The term Federated Learning was coined as recently as 2016 to describe a machine learning setting where multiple entities collaborate in solving a machine learning problem, under the coordination of a central server or service provider. Each client’s raw data is stored locally and not exchanged or transferred; instead, focused updates intended for immediate aggregation are used to achieve the learning objective. Since then, the topic has gathered much interest across many different disciplines and the realization that solving many of these interdisciplinary problems likely requires not just machine learning but techniques from distributed optimization, cryptography, security, differential privacy, fairness, compressed sensing, systems, information theory, statistics, and more. This monograph has contributions from leading experts across the disciplines, who describe the latest state-of-the art from their perspective. These contributions have been carefully curated into a comprehensive treatment that enables the reader to understand the work that has been done and get pointers to where effort is required to solve many of the problems before Federated Learning can become a reality in practical systems. Researchers working in the area of distributed systems will find this monograph an enlightening read that may inspire them to work on the many challenging issues that are outlined. This monograph will get the reader up to speed quickly and easily on what is likely to become an increasingly important topic: Federated Learning.

Software for the frontiers of quantum chemistry: An overview of developments in the Q-Chem 5 package

Abstract: This article summarizes technical advances contained in the fifth major release of the Q-Chem quantum chemistry program package, covering developments since 2015. A comprehensive library of exchange-correlation functionals, along with a suite of correlated many-body methods, continues to be a hallmark of the Q-Chem software. The many-body methods include novel variants of both coupled-cluster and configuration-interaction approaches along with methods based on the algebraic diagrammatic construction and variational reduced density-matrix methods. Methods highlighted in Q-Chem 5 include a suite of tools for modeling core-level spectroscopy, methods for describing metastable resonances, methods for computing vibronic spectra, the nuclear-electronic orbital method, and several different energy decomposition analysis techniques. High-performance capabilities including multithreaded parallelism and support for calculations on graphics processing units are described. Q-Chem boasts a community of well over 100 active academic developers, and the continuing evolution of the software is supported by an "open teamware" model and an increasingly modular design.

End-to-end privacy preserving deep learning on multi-institutional medical imaging

Abstract: Using large, multi-national datasets for high-performance medical imaging AI systems requires innovation in privacy-preserving machine learning so models can train on sensitive data without requiring data transfer. Here we present PriMIA (Privacy-preserving Medical Image Analysis), a free, open-source software framework for differentially private, securely aggregated federated learning and encrypted inference on medical imaging data. We test PriMIA using a real-life case study in which an expert-level deep convolutional neural network classifies paediatric chest X-rays; the resulting model’s classification performance is on par with locally, non-securely trained models. We theoretically and empirically evaluate our framework’s performance and privacy guarantees, and demonstrate that the protections provided prevent the reconstruction of usable data by a gradient-based model inversion attack. Finally, we successfully employ the trained model in an end-to-end encrypted remote inference scenario using secure multi-party computation to prevent the disclosure of the data and the model. Gaining access to medical data to train AI applications can present problems due to patient privacy or proprietary interests. A way forward can be privacy-preserving federated learning schemes. Kaissis, Ziller and colleagues demonstrate here their open source framework for privacy-preserving medical image analysis in a remote inference scenario.

Attention Based Vehicle Trajectory Prediction

Abstract: Self-driving vehicles need to continuously analyse the driving scene, understand the behavior of other road users and predict their future trajectories in order to plan a safe motion and reduce their reaction time. Motivated by this idea, this paper addresses the problem of vehicle trajectory prediction over an extended horizon. On highways, human drivers continuously adapt their speed and paths according to the behavior of their neighboring vehicles. Therefore, vehicles’ trajectories are very correlated and considering vehicle interactions makes motion prediction possible even before the start of a clear maneuver pattern. To this end, we introduce and analyze trajectory prediction methods based on how they model the vehicles interactions. Inspired by human reasoning, we use an attention mechanism that explicitly highlights the importance of neighboring vehicles with respect to their future states. We go beyond pairwise vehicle interactions and model higher order interactions. Moreover, the existence of different goals and driving behaviors induces multiple potential futures. We exploit a combination of global and partial attention paid to surrounding vehicles to generate different possible trajectory. Experiments on highway datasets show that the proposed model outperforms the state-of-the-art performances.

A Test of the Cosmological Principle with Quasars

Abstract: We study the large-scale anisotropy of the universe by measuring the dipole in the angular distribution of a flux-limited, all-sky sample of 1.36 million quasars observed by the Wide-field Infrared Survey Explorer (WISE). This sample is derived from the new CatWISE2020 catalog, which contains deep photometric measurements at 3.4 and 4.6 μm from the cryogenic, post-cryogenic, and reactivation phases of the WISE mission. While the direction of the dipole in the quasar sky is similar to that of the cosmic microwave background (CMB), its amplitude is over twice as large as expected, rejecting the canonical, exclusively kinematic interpretation of the CMB dipole with a p-value of 5 × 10−7 (4.9σ for a normal distribution, one-sided), the highest significance achieved to date in such studies. Our results are in conflict with the cosmological principle, a foundational assumption of the concordance ΛCDM model.

Uncovering the structure of clinical EEG signals with self-supervised learning.

Abstract: Objective. Supervised learning paradigms are often limited by the amount of labeled data that is available. This phenomenon is particularly problematic in clinically-relevant data, such as electroencephalography (EEG), where labeling can be costly in terms of specialized expertise and human processing time. Consequently, deep learning architectures designed to learn on EEG data have yielded relatively shallow models and performances at best similar to those of traditional feature-based approaches. However, in most situations, unlabeled data is available in abundance. By extracting information from this unlabeled data, it might be possible to reach competitive performance with deep neural networks despite limited access to labels. Approach. We investigated self-supervised learning (SSL), a promising technique for discovering structure in unlabeled data, to learn representations of EEG signals. Specifically, we explored two tasks based on temporal context prediction as well as contrastive predictive coding on two clinically-relevant problems: EEG-based sleep staging and pathology detection. We conducted experiments on two large public datasets with thousands of recordings and performed baseline comparisons with purely supervised and hand-engineered approaches. Main results. Linear classifiers trained on SSL-learned features consistently outperformed purely supervised deep neural networks in low-labeled data regimes while reaching competitive performance when all labels were available. Additionally, the embeddings learned with each method revealed clear latent structures related to physiological and clinical phenomena, such as age effects. Significance. We demonstrate the benefit of SSL approaches on EEG data. Our results suggest that self-supervision may pave the way to a wider use of deep learning models on EEG data.

Explainability in deep reinforcement learning

Abstract: A large set of the explainable Artificial Intelligence (XAI) literature is emerging on feature relevance techniques to explain a deep neural network (DNN) output or explaining models that ingest image source data. However, assessing how XAI techniques can help understand models beyond classification tasks, e.g. for reinforcement learning (RL), has not been extensively studied. We review recent works in the direction to attain Explainable Reinforcement Learning (XRL), a relatively new subfield of Explainable Artificial Intelligence, intended to be used in general public applications, with diverse audiences, requiring ethical, responsible and trustable algorithms. In critical situations where it is essential to justify and explain the agent’s behaviour, better explainability and interpretability of RL models could help gain scientific insight on the inner workings of what is still considered a black box. We evaluate mainly studies directly linking explainability to RL, and split these into two categories according to the way the explanations are generated: transparent algorithms and post-hoc explainability. We also review the most prominent XAI works from the lenses of how they could potentially enlighten the further deployment of the latest advances in RL, in the demanding present and future of everyday problems.

Safety and immunogenicity of a two-dose heterologous Ad26.ZEBOV and MVA-BN-Filo Ebola vaccine regimen in adults in Europe (EBOVAC2): a randomised, observer-blind, participant-blind, placebo-controlled, phase 2 trial.

Abstract: Summary Background To address the unmet medical need for an effective prophylactic vaccine against Ebola virus we assessed the safety and immunogenicity of three different two-dose heterologous vaccination regimens with a replication-deficient adenovirus type 26 vector-based vaccine (Ad26.ZEBOV), expressing Zaire Ebola virus glycoprotein, and a non-replicating, recombinant, modified vaccinia Ankara (MVA) vector-based vaccine, encoding glycoproteins from Zaire Ebola virus, Sudan virus, and Marburg virus, and nucleoprotein from the Tai Forest virus. Methods This randomised, observer-blind, placebo-controlled, phase 2 trial was done at seven hospitals in France and two research centres in the UK. Healthy adults (aged 18–65 years) with no history of Ebola vaccination were enrolled into four cohorts. Participants in cohorts I–III were randomly assigned (1:1:1) using computer-generated randomisation codes into three parallel groups (randomisation for cohorts II and III was stratified by country and age), in which participants were to receive an intramuscular injection of Ad26.ZEBOV on day 1, followed by intramuscular injection of MVA-BN-Filo at either 28 days (28-day interval group), 56 days (56-day interval group), or 84 days (84-day interval group) after the first vaccine. Within these three groups, participants in cohort II (14:1) and cohort III (10:3) were further randomly assigned to receive either Ad26.ZEBOV or placebo on day 1, followed by either MVA-BN-Filo or placebo on days 28, 56, or 84. Participants in cohort IV were randomly assigned (5:1) to receive one dose of either Ad26.ZEBOV or placebo on day 1 for vector shedding assessments. For cohorts II and III, study site personnel, sponsor personnel, and participants were masked to vaccine allocation until all participants in these cohorts had completed the post-MVA-BN-Filo vaccination visit at 6 months or had discontinued the trial, whereas cohort I was open-label. For cohort IV, study site personnel and participants were masked to vaccine allocation until all participants in this cohort had completed the post-vaccination visit at 28 days or had discontinued the trial. The primary outcome, analysed in all participants who had received at least one dose of vaccine or placebo (full analysis set), was the safety and tolerability of the three vaccination regimens, as assessed by participant-reported solicited local and systemic adverse events within 7 days of receiving both vaccines, unsolicited adverse events within 42 days of receiving the MVA-BN-Filo vaccine, and serious adverse events over 365 days of follow-up. The secondary outcome was humoral immunogenicity, as measured by the concentration of Ebola virus glycoprotein-binding antibodies at 21 days after receiving the MVA-BN-Filo vaccine. The secondary outcome was assessed in the per-protocol analysis set. This study is registered at ClinicalTrials.gov , NCT02416453 , and EudraCT, 2015-000596-27. Findings Between June 23, 2015, and April 27, 2016, 423 participants were enrolled: 408 in cohorts I–III were randomly assigned to the 28-day interval group (123 to receive Ad26.ZEBOV and MVA-BN-Filo, and 13 to receive placebo), the 56-day interval group (124 to receive Ad26.ZEBOV and MVA-BN-Filo, and 13 to receive placebo), and the 84-day interval group (117 to receive Ad26.ZEBOV and MVA-BN-Filo, and 18 to receive placebo), and 15 participants in cohort IV were assigned to receive Ad26.ZEBOV and MVA-BN-Filo (n=13) or to receive placebo (n=2). 421 (99·5%) participants received at least one dose of vaccine or placebo. The trial was temporarily suspended after two serious neurological adverse events were reported, one of which was considered as possibly related to vaccination, and per-protocol vaccination was disrupted for some participants. Vaccinations were generally well tolerated. Mild or moderate local adverse events (mostly pain) were reported after 206 (62%) of 332 Ad26.ZEBOV vaccinations, 136 (58%) of 236 MVA-BN-Filo vaccinations, and 11 (15%) of 72 placebo injections. Systemic adverse events were reported after 255 (77%) Ad26.ZEBOV vaccinations, 116 (49%) MVA-BN-Filo vaccinations, and 33 (46%) placebo injections, and included mostly mild or moderate fatigue, headache, or myalgia. Unsolicited adverse events occurred after 115 (35%) of 332 Ad26.ZEBOV vaccinations, 81 (34%) of 236 MVA-BN-Filo vaccinations, and 24 (33%) of 72 placebo injections. At 21 days after receiving the MVA-BN-Filo vaccine, geometric mean concentrations of Ebola virus glycoprotein-binding antibodies were 4627 ELISA units (EU)/mL (95% CI 3649–5867) in the 28-day interval group, 10 131 EU/mL (8554–11 999) in the 56-day interval group, and 11 312 mL (9072–14106) in the 84-day interval group, with antibody concentrations persisting at 1149–1205 EU/mL up to day 365. Interpretation The two-dose heterologous regimen with Ad26.ZEBOV and MVA-BN-Filo was safe, well tolerated, and immunogenic, with humoral and cellular immune responses persisting for 1 year after vaccination. Taken together, these data support the intended prophylactic indication for the vaccine regimen. Funding Innovative Medicines Initiative and Janssen Vaccines & Prevention BV. Translation For the French translation of the abstract see Supplementary Materials section.

The JOREK non-linear extended MHD code and applications to large-scale instabilities and their control in magnetically confined fusion plasmas

Abstract: JOREK is a massively parallel fully implicit non-linear extended magneto-hydrodynamic (MHD) code for realistic tokamak X-point plasmas. It has become a widely used versatile simulation code for studying large-scale plasma instabilities and their control and is continuously developed in an international community with strong involvements in the European fusion research programme and ITER organization. This article gives a comprehensive overview of the physics models implemented, numerical methods applied for solving the equations and physics studies performed with the code. A dedicated section highlights some of the verification work done for the code. A hierarchy of different physics models is available including a free boundary and resistive wall extension and hybrid kinetic-fluid models. The code allows for flux-surface aligned iso-parametric finite element grids in single and double X-point plasmas which can be extended to the true physical walls and uses a robust fully implicit time stepping. Particular focus is laid on plasma edge and scrape-off layer (SOL) physics as well as disruption related phenomena. Among the key results obtained with JOREK regarding plasma edge and SOL, are deep insights into the dynamics of edge localized modes (ELMs), ELM cycles, and ELM control by resonant magnetic perturbations, pellet injection, as well as by vertical magnetic kicks. Also ELM free regimes, detachment physics, the generation and transport of impurities during an ELM, and electrostatic turbulence in the pedestal region are investigated. Regarding disruptions, the focus is on the dynamics of the thermal quench (TQ) and current quench triggered by massive gas injection and shattered pellet injection, runaway electron (RE) dynamics as well as the RE interaction with MHD modes, and vertical displacement events. Also the seeding and suppression of tearing modes (TMs), the dynamics of naturally occurring TQs triggered by locked modes, and radiative collapses are being studied.

Thinking Fast and Slow: Efficient Text-to-Visual Retrieval with Transformers

Abstract: Our objective is language-based search of large-scale image and video datasets. For this task, the approach that consists of independently mapping text and vision to a joint embedding space, a.k.a. dual encoders, is attractive as retrieval scales and is efficient for billions of images using approximate nearest neighbour search. An alternative approach of using vision-text transformers with cross-attention gives considerable improvements in accuracy over the joint embeddings, but is often inapplicable in practice for large-scale retrieval given the cost of the cross-attention mechanisms required for each sample at test time. This work combines the best of both worlds. We make the following three contributions. First, we equip transformer-based models with a new fine-grained cross-attention architecture, providing significant improvements in retrieval accuracy whilst preserving scalability. Second, we introduce a generic approach for combining a Fast dual encoder model with our Slow but accurate transformer-based model via distillation and reranking. Finally, we validate our approach on the Flickr30K image dataset where we show an increase in inference speed by several orders of magnitude while having results competitive to the state of the art. We also extend our method to the video domain, improving the state of the art on the VATEX dataset.

When Being Unseen from mBERT is just the Beginning: Handling New Languages With Multilingual Language Models

Abstract: Transfer learning based on pretraining language models on a large amount of raw data has become a new norm to reach state-of-the-art performance in NLP. Still, it remains unclear how this approach should be applied for unseen languages that are not covered by any available large-scale multilingual language model and for which only a small amount of raw data is generally available. In this work, by comparing multilingual and monolingual models, we show that such models behave in multiple ways on unseen languages. Some languages greatly benefit from transfer learning and behave similarly to closely related high resource languages whereas others apparently do not. Focusing on the latter, we show that this failure to transfer is largely related to the impact of the script used to write such languages. We show that transliterating those languages significantly improves the potential of large-scale multilingual language models on downstream tasks. This result provides a promising direction towards making these massively multilingual models useful for a new set of unseen languages.

An Introduction to Topological Data Analysis: Fundamental and Practical Aspects for Data Scientists

Abstract: Topological Data Analysis (TDA)is a recent and fast growing field providing a set of new topological and geometric tools to infer relevant features for possibly complex data. This paper is a brief introduction, through a few selected topics, to basic fundamental and practical aspects of TDA for non experts.

Fasting alters the gut microbiome reducing blood pressure and body weight in metabolic syndrome patients.

Abstract: Periods of fasting and refeeding may reduce cardiometabolic risk elevated by Western diet. Here we show in the substudy of NCT02099968, investigating the clinical parameters, the immunome and gut microbiome exploratory endpoints, that in hypertensive metabolic syndrome patients, a 5-day fast followed by a modified Dietary Approach to Stop Hypertension diet reduces systolic blood pressure, need for antihypertensive medications, body-mass index at three months post intervention compared to a modified Dietary Approach to Stop Hypertension diet alone. Fasting alters the gut microbiome, impacting bacterial taxa and gene modules associated with short-chain fatty acid production. Cross-system analyses reveal a positive correlation of circulating mucosa-associated invariant T cells, non-classical monocytes and CD4+ effector T cells with systolic blood pressure. Furthermore, regulatory T cells positively correlate with body-mass index and weight. Machine learning analysis of baseline immunome or microbiome data predicts sustained systolic blood pressure response within the fasting group, identifying CD8+ effector T cells, Th17 cells and regulatory T cells or Desulfovibrionaceae, Hydrogenoanaerobacterium, Akkermansia, and Ruminococcaceae as important contributors to the model. Here we report that the high-resolution multi-omics data highlight fasting as a promising non-pharmacological intervention for the treatment of high blood pressure in metabolic syndrome patients.

The gem5 Simulator: Version 20.0+

Abstract: The open-source and community-supported gem5 simulator is one of the most popular tools for computer architecture research. This simulation infrastructure allows researchers to model modern computer hardware at the cycle level, and it has enough fidelity to boot unmodified Linux-based operating systems and run full applications for multiple architectures including x86, Arm®, and RISC-V. The gem5 simulator has been under active development over the last nine years since the original gem5 release. In this time, there have been over 7000 commits to the codebase from over 250 unique contributors which have improved the simulator by adding new features, fixing bugs, and increasing the code quality. In this paper, we give an overview of gem5's usage and features, describe the current state of the gem5 simulator, and enumerate the major changes since the initial release of gem5. We also discuss how the gem5 simulator has transitioned to a formal governance model to enable continued improvement and community support for the next 20 years of computer architecture research.

PySyft: A Library for Easy Federated Learning

Abstract: PySyft is an open-source multi-language library enabling secure and private machine learning by wrapping and extending popular deep learning frameworks such as PyTorch in a transparent, lightweight, and user-friendly manner. Its aim is to both help popularize privacy-preserving techniques in machine learning by making them as accessible as possible via Python bindings and common tools familiar to researchers and data scientists, as well as to be extensible such that new Federated Learning (FL), Multi-Party Computation, or Differential Privacy methods can be flexibly and simply implemented and integrated. This chapter will introduce the methods available within the PySyft library and describe their implementations. We will then provide a proof-of-concept demonstration of a FL workflow using an example of how to train a convolutional neural network. Next, we review the use of PySyft in academic literature to date and discuss future use-cases and development plans. Most importantly, we introduce Duet: our tool for easier FL for scientists and data owners.

How can we combat multicenter variability in MR radiomics? Validation of a correction procedure

Abstract: Test a practical realignment approach to compensate the technical variability of MR radiomic features. T1 phantom images acquired on 2 scanners, FLAIR and contrast-enhanced T1-weighted (CE-T1w) images of 18 brain tumor patients scanned on both 1.5-T and 3-T scanners, and 36 T2-weighted (T2w) images of prostate cancer patients scanned in one of two centers were investigated. The ComBat procedure was used for harmonizing radiomic features. Differences in statistical distributions in feature values between 1.5- and 3-T images were tested before and after harmonization. The prostate studies were used to determine the impact of harmonization to distinguish between Gleason grades (GGs). In the phantom data, 40 out of 42 radiomic feature values were significantly different between the 2 scanners before harmonization and none after. In white matter regions, the statistical distributions of features were significantly different (p < 0.05) between the 1.5- and 3-T images for 37 out of 42 features in both FLAIR and CE-T1w images. After harmonization, no statistically significant differences were observed. In brain tumors, 41 (FLAIR) or 36 (CE-T1w) out of 42 features were significantly different between the 1.5- and 3-T images without harmonization, against 1 (FLAIR) or none (CE-T1w) with harmonization. In prostate studies, 636 radiomic features were significantly different between GGs after harmonization against 461 before. The ability to distinguish between GGs using radiomic features was increased after harmonization. ComBat harmonization efficiently removes inter-center technical inconsistencies in radiomic feature values and increases the sensitivity of studies using data from several scanners. • Radiomic feature values obtained using different MR scanners or imaging protocols can be harmonized by combining off-the-shelf image standardization and feature realignment procedures. • Harmonized radiomic features enable one to pool data from different scanners and centers without a substantial loss of statistical power caused by intra- and inter-center variability. • The proposed realignment method is applicable to radiomic features from different MR sequences and tumor types and does not rely on any phantom acquisition.

SoK: Computer-Aided Cryptography

Abstract: Computer-aided cryptography is an active area of research that develops and applies formal, machine-checkable approaches to the design, analysis, and implementation of cryptography. We present a cross-cutting systematization of the computer-aided cryptography literature, focusing on three main areas: (i) design-level security (both symbolic security and computational security), (ii) functional correctness and efficiency, and (iii) implementation-level security (with a focus on digital side-channel resistance). In each area, we first clarify the role of computer-aided cryptography—how it can help and what the caveats are—in addressing current challenges. We next present a taxonomy of state-of-the-art tools, comparing their accuracy, scope, trustworthiness, and usability. Then, we highlight their main achievements, trade-offs, and research challenges. After covering the three main areas, we present two case studies. First, we study efforts in combining tools focused on different areas to consolidate the guarantees they can provide. Second, we distill the lessons learned from the computer-aided cryptography community’s involvement in the TLS 1.3 standardization effort. Finally, we conclude with recommendations to paper authors, tool developers, and standardization bodies moving forward.

Deformable Kernel Networks for Joint Image Filtering

Abstract: Joint image filters are used to transfer structural details from a guidance picture used as a prior to a target image, in tasks such as enhancing spatial resolution and suppressing noise. Previous methods based on convolutional neural networks (CNNs) combine nonlinear activations of spatially-invariant kernels to estimate structural details and regress the filtering result. In this paper, we instead learn explicitly sparse and spatially-variant kernels. We propose a CNN architecture and its efficient implementation, called the deformable kernel network (DKN), that outputs sets of neighbors and the corresponding weights adaptively for each pixel. The filtering result is then computed as a weighted average. We also propose a fast version of DKN that runs about seventeen times faster for an image of size $$640 \times 480$$ . We demonstrate the effectiveness and flexibility of our models on the tasks of depth map upsampling, saliency map upsampling, cross-modality image restoration, texture removal, and semantic segmentation. In particular, we show that the weighted averaging process with sparsely sampled $$3 \times 3$$ kernels outperforms the state of the art by a significant margin in all cases.

On the use of simulation in robotics: Opportunities, challenges, and suggestions for moving forward.

Abstract: The last five years marked a surge in interest for and use of smart robots, which operate in dynamic and unstructured environments and might interact with humans. We posit that well-validated computer simulation can provide a virtual proving ground that in many cases is instrumental in understanding safely, faster, at lower costs, and more thoroughly how the robots of the future should be designed and controlled for safe operation and improved performance. Against this backdrop, we discuss how simulation can help in robotics, barriers that currently prevent its broad adoption, and potential steps that can eliminate some of these barriers. The points and recommendations made concern the following simulation-in-robotics aspects: simulation of the dynamics of the robot; simulation of the virtual world; simulation of the sensing of this virtual world; simulation of the interaction between the human and the robot; and, in less depth, simulation of the communication between robots. This Perspectives contribution summarizes the points of view that coalesced during a 2018 National Science Foundation/Department of Defense/National Institute for Standards and Technology workshop dedicated to the topic at hand. The meeting brought together participants from a range of organizations, disciplines, and application fields, with expertise at the intersection of robotics, machine learning, and physics-based simulation.

A review of user training methods in brain computer interfaces based on mental tasks.

Abstract: Mental-Tasks based Brain-Computer Interfaces (MT-BCIs) allow their users to interact with an external device solely by using brain signals produced through mental tasks. While MT-BCIs are promising for many applications, they are still barely used outside laboratories due to their lack of reliability. MT-BCIs require their users to develop the ability to self-regulate specific brain signals. However, the human learning process to control a BCI is still relatively poorly understood and how to optimally train this ability is currently under investigation. Despite their promises and achievements, traditional training programs have been shown to be sub-optimal and could be further improved. In order to optimize user training and improve BCI performance, human factors should be taken into account. An interdisciplinary approach should be adopted to provide learners with appropriate and/or adaptive training. In this article, we provide an overview of existing methods for MT-BCI user training - notably in terms of environment, instructions, feedback and exercises. We present a categorization and taxonomy of these training approaches, provide guidelines on how to choose the best methods and identify open challenges and perspectives to further improve MT-BCI user training.

Hardware Private Circuits: From Trivial Composition to Full Verification

Abstract: The design of glitch-resistant higher-order masking schemes is an important challenge in cryptographic engineering. A recent work by Moos et al. (CHES 2019) showed that most published schemes (and all efficient ones) exhibit local or composability flaws at high security orders, leaving a critical gap in the literature on hardware masking. In this article, we first extend the simulatability framework of Belaid et al. (EUROCRYPT 2016) and prove that a compositional strategy that is correct without glitches remains valid with glitches. We then use this extended framework to prove the first masked gadgets that enable trivial composition with glitches at arbitrary orders. We show that the resulting “Hardware Private Circuits” approach the implementation efficiency of previous (flawed) schemes. We finally investigate how trivial composition can serve as a basis for a tool that allows verifying full masked hardware implementations (e.g., of complete block ciphers) at any security order from their HDL code. As side products, we improve the randomness complexity of the best published refreshing gadgets, show that some S-box representations allow latency reductions and confirm practical claims based on implementation results.

Analyzing Dynamic Hypergraphs with Parallel Aggregated Ordered Hypergraph Visualization

Abstract: Parallel Aggregated Ordered Hypergraph(PAOH) is a novel technique to visualize dynamic hypergraphs. Hypergraphs are a generalization of graphs where edges can connect several vertices. Hypergraphs can be used to model networks of business partners or co-authorship networks with multiple authors per article. A dynamic hypergraph evolves over discrete time slots. PAOH represents vertices as parallel horizontal bars and hyperedges as vertical lines, using dots to depict the connections to one or more vertices. We describe a prototype implementation of Parallel Aggregated Ordered Hypergraph, report on a usability study with 9 participants analyzing publication data, and summarize the improvements made. Two case studies and several examples are provided. We believe that PAOH is the first technique to provide a highly readable representation of dynamic hypergraphs. It is easy to learn and well suited for medium size dynamic hypergraphs (50-500 vertices) such as those commonly generated by digital humanities projects—our driving application domain.

The Surfactin-Like Lipopeptides From Bacillus spp.: Natural Biodiversity and Synthetic Biology for a Broader Application Range.

Abstract: Surfactin is a lipoheptapeptide produced by several Bacillus species and identified for the first time in 1969. At first, the biosynthesis of this remarkable biosurfactant was described in this review. The peptide moiety of the surfactin is synthesized using huge multienzymatic proteins called NonRibosomal Peptide Synthetases. This mechanism is responsible for the peptide biodiversity of the members of the surfactin family. In addition, on the fatty acid side, fifteen different isoforms (from C12 to C17) can be incorporated so increasing the number of the surfactin-like biomolecules. The review also highlights the last development in metabolic modeling and engineering and in synthetic biology to direct surfactin biosynthesis but also to generate novel derivatives. This large set of different biomolecules leads to a broad spectrum of physico-chemical properties and biological activities. The last parts of the review summarized the numerous studies related to the production processes optimization as well as the approaches developed to increase the surfactin productivity of Bacillus cells taking into account the different steps of its biosynthesis from gene transcription to surfactin degradation in the culture medium.

Benchmarking joint multi-omics dimensionality reduction approaches for the study of cancer.

Abstract: High-dimensional multi-omics data are now standard in biology. They can greatly enhance our understanding of biological systems when effectively integrated. To achieve proper integration, joint Dimensionality Reduction (jDR) methods are among the most efficient approaches. However, several jDR methods are available, urging the need for a comprehensive benchmark with practical guidelines. We perform a systematic evaluation of nine representative jDR methods using three complementary benchmarks. First, we evaluate their performances in retrieving ground-truth sample clustering from simulated multi-omics datasets. Second, we use TCGA cancer data to assess their strengths in predicting survival, clinical annotations and known pathways/biological processes. Finally, we assess their classification of multi-omics single-cell data. From these in-depth comparisons, we observe that intNMF performs best in clustering, while MCIA offers an effective behavior across many contexts. The code developed for this benchmark study is implemented in a Jupyter notebook-multi-omics mix (momix)-to foster reproducibility, and support users and future developers.

Fullsubnet: A Full-Band and Sub-Band Fusion Model for Real-Time Single-Channel Speech Enhancement

Abstract: This paper proposes a full-band and sub-band fusion model, named as FullSubNet, for single-channel real-time speech enhancement. Full-band and sub-band refer to the models that input full-band and sub-band noisy spectral feature, output full-band and sub-band speech target, respectively. The sub-band model processes each frequency independently. Its input consists of one frequency and several context frequencies. The output is the prediction of the clean speech target for the corresponding frequency. These two types of models have distinct characteristics. The full-band model can capture the global spectral context and the long-distance cross-band dependencies. However, it lacks the ability to modeling signal stationarity and attending the local spectral pattern. The sub-band model is just the opposite. In our proposed FullSubNet, we connect a pure full-band model and a pure sub-band model sequentially and use practical joint training to integrate these two types of models’ advantages. We conducted experiments on the DNS challenge (INTERSPEECH 2020) dataset to evaluate the proposed method. Experimental results show that full-band and sub-band information are complementary, and the FullSubNet can effectively integrate them. Besides, the performance of the FullSubNet also exceeds that of the top-ranked methods in the DNS Challenge (INTERSPEECH 2020).