Microsoft Windows | EPFL Graph Search

Microsoft Windows is a group of several proprietary graphical operating system families developed and marketed by Microsoft. Each family caters to a certain sector of the computing industry. For example, Windows NT for consumers, Windows Server for servers, and Windows IoT for embedded systems. Defunct Windows families include Windows 9x, Windows Mobile, and Windows Phone. The first version of Windows was released on November 20, 1985, as a graphical operating system shell for MS-DOS in response to the growing interest in graphical user interfaces (GUIs). Windows is the most popular desktop operating system in the world, with a 70% market share , according to StatCounter. However, Windows is not the most used operating system when including both mobile and desktop OSes, due to Android's massive growth. , the most recent version of Windows is Windows 11 for consumer PCs and tablets, Windows 11 Enterprise for corporations, and Windows Server 2022 for servers. Genealogy

Problèmes de cheminements optimaux dans les réseaux avec contraintes associées aux arcs

In this thesis we consider two optimal routing problems in networks. Both are NP-hard and are often used to modelize real life problems of large size. We first present the capacitated arc routing problem (CARP). In this problem, we have to serve a set of customers represented by the arcs of a network with vehicles of restricted capacity. The aim is to design a set of vehicle routes of least total cost. We describe two heuristic methods for the CARP based on local search techniques. The first one is a Tabu Search heuristic using an original post-optimization procedure. Principles appearing in this procedure are used to define a class of neighborhoods we combine in a Variable Neighborhood Descent method. Our algorithms compare favorably with some of her methods. They were initially developed to deal with the undirected CARP. We describe how to transform them so that they can be applied to the directed CARP. We also propose an adaptation to the directed case of a lower bound originally developed for the undirected CARP. This lower bound is used to evaluate the quality of the solutions produced by our algorithms for the directed CARP. Our methods obtain solutions whose value is, on average, close to known lower bounds. We use the CARP to tackle a real life problem of boat routing and scheduling. This problem can be viewed as a directed CARP with time windows. We describe in detail its modeling process and we analyze our first results. We obtain solutions that require less boats than the present schedule. Nevertheless, these solutions are not completely feasible in practice. We suggest some ways to improve them. The second problem treated in this thesis is the shortest capacitated paths problem (SCPP). We present a Tabu Search heuristic and some post-optimization procedures for the SCPP. These procedures are called in our Tabu Search algorithm. Two versions of this algorithm are considered. In the first one, post-optimization procedures are intensively used. In the second one, they are less frequently applied. This second version is faster. Compared with a greedy approach, the two versions of our Tabu Search heuristic obtain solutions which are, on average, of better quality. The SCPP was proposed to us by EDF (Electricité de France) in order to minimize the total length of the paths followed by cables in a nuclear power-station. It can also be viewed as a subproblem of the bandwidth packing problem (BWP) appearing in telecommunications.

EPFL2000

Predictive Analysis of Errors During Robot-Mediated Gamified Training

Pierre Dillenbourg, Nevena Dresevic, Arzu Güneysu Özgür, Marija Katanic, Nihal Ezgi Yücetürk

This paper presents our approach to predicting future error-related events in a robot-mediated gamified phys- ical training activity for stroke patients. The ability to predict future error under such conditions suggests the existence of distinguishable features and separated class characteristics between the casual gameplay state and error prune state in the data. Identifying such features provides valuable insight to creating individually tailored, adaptive games as well as possible ways to increase rehabilitation success by patients. Considering the time-series nature of sensory data created by motor actions of patients we employed a predictive analysis strategy on carefully engineered features of sequenced data. We split the data into fixed time windows and explored logistic regression models, decision trees, and recurrent neural networks to predict the likelihood of a patient making an error based on the features from the time window before the error. We achieved an 84.4% F1-score with a 0.76 ROC value in our best model for predicting motion accuracy related errors. Moreover, we computed the permutation importance of the features to explain which ones are more indicative of future errors.

2022

Self-attention for Speech Emotion Recognition

Philip Neil Garner, Lorenzo Tarantino

Speech Emotion Recognition (SER) has been shown to benefit from many of the recent advances in deep learning, including recurrent based and attention based neural network architectures as well. Nevertheless, performance still falls short of that of humans. In this work, we investigate whether SER could benefit from the self-attention and global windowing of the transformer model. We show on the IEMOCAP database that this is indeed the case. Finally, we investigate whether using the distribution of, possibly conflicting, annotations in the training data, as soft targets could outperform a majority voting. We prove that this performance increases with the agreement level of the annotators.

2019

Problèmes de cheminements optimaux dans les réseaux avec contraintes associées aux arcs

EPFL2000