International Professional University of Technology in Nagoya Repository
Not a member yet
15131 research outputs found
Sort by
Application of sequential pattern mining to the analysis of visitor trajectories
No full textInternational audienceIn this work, we demonstrate the proof of concept of clustering 254 visitors based on their trajectories in a museum. We used a real dataset from Haifa Museum, where each trajectory is treated as a sequence of itemsets. We applied simACS as a similarity measure between any two sequences
A global optimization approach for rational sparsity promoting criteria
No full textInternational audienceWe consider the problem of recovering an unknown signal observed through a nonlinear model and corrupted with additive noise. More precisely, the nonlinear degradation consists of a convolution followed by a nonlinear rational transform. As a prior information, the original signal is assumed to be sparse. We tackle the problem by minimizing a least-squares fit criterion penalized by a Geman-McClure like potential. In order to find a globally optimal solution to this rational minimization problem, we transform it in a generalized moment problem, for which a hierarchy of semidefinite programming relaxations can be used. To overcome computational limitations on the number of involved variables, the structure of the problem is carefully addressed, yielding a sparse relaxation able to deal with up to several hundreds of optimized variables. Our experiments show the good performance of the proposed approac
Summary of the Weizmann workshop: Hammers & Nails - Machine Learning & HEP
No full textInternational audienc
Performance Analysis of Cluster Formation in Wireless Sensor Networks
No full textInternational audienceClustered-based wireless sensor networks have been extensively used in the literature in order to achieve considerable energy consumption reductions. However, two aspects of such systems have been largely overlooked. Namely, the transmission probability used during the cluster formation phase and the way in which cluster heads are selected. Both of these issues have an important impact on the performance of the system. For the former, it is common to consider that sensor nodes in a clustered-based Wireless Sensor Network (WSN) use a fixed transmission probability to send control data in order to build the clusters. However, due to the highly variable conditions experienced by these networks, a fixed transmission probability may lead to extra energy consumption. In view of this, three different transmission probability strategies are studied: optimal, fixed and adaptive. In this context, we also investigate cluster head selection schemes, specifically, we consider two intelligent schemes based on the fuzzy C-means and k-medoids algorithms and a random selection with no intelligence. We show that the use of intelligent schemes greatly improves the performance of the system, but their use entails higher complexity and selection delay. The main performance metrics considered in this work are energy consumption , successful transmission probability and cluster formation latency. As an additional feature of this work, we study the effect of errors in the wireless channel and the impact on the performance of the system under the different transmission probability schemes
Le défi des 1001 graphes
No full textNational audienceDans un graphe, existe-t-il un circuit visitant chaque sommet une fois et une seule ? Une question difficile pour certains graphes...Cet article explique comment les auteurs ont abordé et remporté la compétition internationale organisée par la Flinders University d'Adelaïde (Australie), intitulée FHCP Challenge, sur le problème du cycle hamiltonien dans les graphes
Modélisation mathématique pour l'étude des oscillations neuronales dans des réseaux de mémoire hippocampiques pendant l'éveil et sous anesthésie générale
No full textMemory is commonly defined as the ability to encode, store, and recall information we perceived. As we experience the world, we sense stimuli, we witness events, we ascertain facts, we study concepts, and we acquire skills. Although memory is an innate and familiar human behaviour, the interior workings of the brain which provide us with such faculties are far from being fully unravelled. Experimental studies have shown that during memory tasks, certain brain structures exhibit synchronous activity which is thought to be correlated with the short-term maintenance of salient stimuli. The objective of this thesis is to use biologically-inspired mathematical modelling and simulations of neural activity to shed some light on the mechanisms enabling the emergence of these memory-related synchronous oscillations. We focus in particular on hippocampal mnemonic activity during the awake state, and the amnesia and paradoxical memory consolidation occurring under general anaesthesia. We begin by introducing a detailed model of a type of persistent-firing pyramidal neuron commonly found in the CA3 and CA1 areas of the hippocampus. Stimulated with a brief transient current pulse, the neuron displays persistent activity maintained solely by cholinergic calcium-activated non-specific (CAN) receptors, and outlasting the stimulus for long delay periods (> 30s). Our model neuron and its parameters are derived from experimental in-vitro recordings of persistent firing hippocampal neurons carried out by our collaborators Beate Knauer and Motoharu Yoshida at the Ruhr University in Bochum, Germany. Subsequently, we turn our attention to the dynamics of a population of such interconnected pyramidal-CAN neurons. We hypothesise that networks of persistent firing neurons could provide the neural mechanism for the maintenance of memory-related hippocampal oscillations. The firing patterns elicited by this network are in accord with both experimental recordings and modelling studies. In addition, the network displays self-sustained oscillatory activity in the theta frequency. When connecting the pyramidal-CAN network to fast-spiking inhibitory interneurons, the dynamics of the model reveal that feedback inhibition improves the robustness of fast theta oscillations, by tightening the synchronisation of the pyramidal CAN neurons. We demonstrate that, in the model, the frequency and spectral power of the oscillations are modulated solely by the cholinergic mechanisms mediating the intrinsic persistent firing, allowing for a wide range of oscillation rates within the theta band. This is a biologically plausible mechanism for the maintenance of synchronous theta oscillations in the hippocampus which aims at extending the traditional models of septum-driven hippocampal rhythmic activity. In addition, we study the disruptive effects of general anaesthesia on hippocampal gamma-frequency oscillations. We present an in-depth study of the action of anaesthesia on neural oscillations by introducing a new computational model which takes into account the four main effects of the anaesthetic agent propofol GABAergic hippocampal interneurons. Our results indicate that propofol-mediated tonic inhibition contributes to enhancing network synchronisation in a network of hippocampal interneurons. This enhanced synchronisation could provide a possible mechanism supporting the occurrence of intraoperative awareness, explicit memory formation, and even paradoxical excitation under general anaesthesia, by facilitating the communication between brain structures which should supposedly be not allowed to do so when anaesthetised. In conclusion, the findings described within this thesis provide new insights into the mechanisms underlying mnemonic neural activity, both during wake and anaesthesia, opening compelling avenues for future work on clinical applications tackling neurodegenerative memory diseases, and anaesthesia monitoringLa mémoire est communément définie comme la capacité de coder, stocker et rappeler les informations que nous avons perçues. Lorsque nous traversons le monde, nous ressentons des stimuli, nous assistons à des événements, nous constatons des faits, nous étudions des concepts et nous acquérons des compétences. Bien que la mémoire soit un comportement humain inné et familier, les mécanismes cérébraux qui nous fournissent de telles facultés sont loin d'être compris. Des études expérimentales ont montré que, lors des tâches de mémoire, certaines structures cérébrales présentent une activité synchrone qui est censée être corrélée avec le maintien à court terme des stimuli saillants. L'objectif de cette thèse est d'utiliser la modélisation mathématique biologiquement inspirée et des simulations d'activité neuronale pour éclairer les mécanismes permettant l'émergence de ces oscillations synchrones liées à la mémoire. Nous nous concentrons en particulier sur l'activité mnémonique de l'hippocampe pendant l'état éveillé, et l'amnésie et la consolidation inattendue de la mémoire sous anesthésie générale. Nous commençons par présenter un modèle détaillé de neurone pyramidal qui se trouve couramment dans les zones CA3 et CA1 de l'hippocampe. Stimulé par une courte impulsion de courant, le neurone produit une activité persistante maintenue pour de longues périodes (> 30s) au-delà du stimulus uniquement par des récepteurs calciques non spécifiques (CAN). Les paramètres du modèle sont dérivés des enregistrements in vitro de neurones hippocampiques réalisés par nos collaborateurs Beate Knauer et Motoharu Yoshida à l'Université de la Ruhr à Bochum, en Allemagne. Par la suite, nous étudions la dynamique d'une population de ces neurones pyramidaux-CAN interconnectés. Nous supposons que les réseaux de neurones à tir persistant pourraient fournir le mécanisme neuronale pour la maintenance des oscillations hippocampiques mnésiques. Nos résultats montrent que le réseau génère une activité oscillante auto-soutenue dans la fréquence thêta. Lors de la connexion du réseau pyramidal-CAN à des interneurones inhibiteurs, la dynamique du modèle révèle que l'inhibition rétroactive améliore la robustesse des oscillations thêta rapides, en resserrant la synchronisation des neurones pyramidaux. Nous démontrons que, dans le modèle, la fréquence et la puissance spectrale des oscillations sont modulées uniquement par le courant CAN, permettant une large gamme de fréquences d'oscillation dans la bande theta. Il s'agit d'un mécanisme biologiquement plausible pour la maintenance des oscillations thêta synchrones dans l'hippocampe qui vise à étendre les modèles traditionnels d'activité rythmique hippocampique entraînée par le septum. En outre, nous présentons une étude approfondie des effets perturbateurs de l'anesthésie générale sur les oscillations gamma dans l'hippocampe. Nous introduisons un nouveau modèle qui prend en compte les quatre principaux effets de l'agent anesthésique propofol sur les récepteurs GABAA. Nos résultats indiquent que l'inhibition tonique médiée par le propofol contribue à améliorer la synchronisation du réseau dans un réseau d'interneurones de l'hippocampe. Cette synchronisation améliorée pourrait fournir une explication possible pour l'apparition d'une conscience intra-opératoire, d'une formation explicite de la mémoire et même d'une excitation paradoxale sous anesthésie générale, en facilitant la communication entre structures cérébrales qui ne devraient pas être autorisées à le faire lorsqu'elles sont anesthésiées. En conclusion, les résultats décrits dans cette thèse fournissent de nouvelles idées sur les mécanismes sous-jacents de l'activité neuronale mnémonique, à la fois au cours du réveil et de l'anesthésie, en ouvrant des voies convaincantes pour les travaux futurs sur les applications cliniques qui s'attaquent aux maladies de la mémoire neurodégénératives et la surveillance de l'anesthésie
Mm-wave antennas and components: Profiting from 3D-printing
No full textInternational audienceThe goal of this manuscript is to show the potential of additive manufacturing (AM) to produce cost-effective high-performance passive mm-wave components and antennas. We focus on the particular AM technique developed by SWISSto12 based on the so called Stereolithography (SLA), which consists on the 3D-printing of a skeleton of the component using a non-conductive polymer. Chemical copper plating is then applied in order to make all component surfaces RF-conductive. One of the most relevant advantages associated to this approach stems from its ability to produce low-weight monolithic devices. Two 3D-printed RF-devices operating at mm-waves are here presented: a Ka-band radiating element and a V-band orthomode-transducer. Both devices have been firstly designed using full-wave solvers, then manufactured and finally validated through measurements. © 2017 IEEE
Recurrent Neural Network-based Fault Detector for Aileron Failures of Aircraft
No full textInternational audienceThis paper empirically investigate the design of a fault detection mechanism based on Long Short Term Memory (LSTM) neural network. Given an equation based model that approximate the behavior of aircraft ailerons, the fault detector aims at predicting the state of aircraft: the normal state for which no failure are observed, or four different failure states, e.g. a delay changes. This is achieved by collecting a limited amount of command and responses data by varying the parameters of the aileron model, such that a LSTM network is used to predict the state of the aircraft of sequence of the pair commands/responses. In this empirical study we empirically demonstrated LSTM networks can be a promising approach for fault detection, and achieve reasonable performances despite a limited amount of data, in particular avoiding overfitting of the model
Multimodal First Impression Analysis with Deep Residual Networks
No full textInternational audiencePeople form first impressions about the personalities of unfamiliar individuals even after very brief interactions with them. In this study we present and evaluate several models that mimic this automatic social behavior. Specifically, we present several models trained on a large dataset of short YouTube video blog posts for predicting apparent Big Five personality traits of people and whether they seem suitable to be recommended to a job interview. Along with presenting our audiovisual approach and results that won the third place in the ChaLearn First Impressions Challenge, we investigate modeling in different modalities including audio only, visual only, language only, audiovisual, and combination of audiovisual and language. Our results demonstrate that the best performance could be obtained using a fusion of all data modalities