Sébastien Aupetit

Maintenance policy: degradation laws versus hidden Markov model availability indicator

Today, maintenance strategies and their analyses remain a worrying problem for companies. Socio-economic stakes depending on the competitiveness of each strategy are more than ever linked to the activity and quality of maintenance interventions. A series of specific events can eventually warn the expert of an imminent breakdown. This study aims at understanding such a signature thanks to hidden Markov models. For that purpose, two methods for damage level estimation of a maintained system are proposed. The first consists in using non-parametric and semi-parametric degradation laws (which will be used as references). The second method consists in using a Markovian approach. All proposals are illustrated on two studies corresponding to two real industrial situations (a continuous system for food processing and moulded products in aluminium alloys for the automotive industry).

Hidden Markov Models Training by a Particle Swarm Optimization Algorithm

In this work we consider the problem of Hidden Markov Models (HMM) training. This problem can be considered as a global optimization problem and we focus our study on the Particle Swarm Optimization (PSO) algorithm. To take advantage of the search strategy adopted by PSO, we need to modify the HMMs search space. Moreover, we introduce a local search technique from the field of HMMs and that is known as the Baum–Welch algorithm. A parameter study is then presented to evaluate the importance of several parameters of PSO on artificial data and natural data extracted from images.

Web Page Textual Color Contrast Compensation for CVD Users Using Optimization Methods

With this paper, we propose two methods for color contrast compensation of the textual information contained in a web page using numerical optimization. The optimization process can be reduced to the minimization of a single objective function which aims to achieve an on-the-fly compensation inducing a small amount of change in the original colors. Mass-spring system based optimization and CMA-ES metaheuristics are compared with the problem in order to assess their efficiency for compensating the loss. Experiments conducted on real and artificial datasets, prove the methods efficiency even with a small number of evaluations. Also, the methods behaviour is bound to the amount of compensation needed.

Improving web accessibility for dichromat users through contrast preservation

Unfortunately, accessibility is not one of designers priorities while developing web sites, resulting in barriers for numerous disabled users. In this context, it is fundamental to identify the difficulties they may experience while surfing web and to propose solutions in order to remove them or diminish their impact. The choice of colors is far from being a random process but often a way to transmit or emphase information. This is particulary true for textual information contained in a web page. The perception of colors by a dichromat user is different. This results in a loss of the information conveyed by color. In our study, we show that there is a significant loss of contrast for a dichromat user resulting in information loss. We propose a method based on a mass-spring simulation to modify the colors with aim to enforce similar contrast for dichromat users. Tests on several websites allow us to conclude that our method significantly reduce the loss of contrast for both protanope and deuteranope users.

Hidden Markov Models Training Using Population-based Metaheuristics

In this chapter, we consider the issue of Hidden Markov Model (HMM) training. First, HMMs are introduced and then we focus on the particular HMM training problem. We emphasize the difficulty of this problem and present various criteria that can be considered.Many different adaptations of metaheuristics have been used but, until now, few extensive comparisons have been performed for this problem. We propose to compare three population-based metaheuristics (genetic algorithm, ant algorithm and particle swarm optimization) with and without the help of a local optimizer. These algorithms make use of solutions that can be explored in three different kinds of search space (a constrained space, a discrete space and a vector space). We study these algorithms from both a theoretical and an experimental perspective: parameter settings are fully studied on a reduced set of data and the performances of algorithms are compared on different sets of real data.

Clustering and dynamic data visualization with artificial flying insect

We present in this paper a new bio-inspired algorithm that dynamically creates and visualizes groups of data. This algorithm uses the concepts of flying insects that move together in complex manner with simple local rules. Each insect represents one datum. The insect moves aim at creating homogeneous groups of data that evolve together in a 2D environment in order to help the domain expert to understand the underlying class structure of the data set.

An Evolutionary Approach to Contrast Compensation for Dichromat Users

In this paper, we are focusing on web accessibility, more precisely on improving web accessibility for Color Vision Deficiency (CVD) users. The contrast optimization problem for dichromat users can be modeled as a mono objective function which at minimum provides a suitable solution to the problem. The function aims to compensate the loss and maintains simultaneously a minimum change in the original color. The CMA-ES method is used to minimize the function. Experiments were conducted on real and artificial data in order to assess the approach efficiency for different set of parameters. The results showed that it is likely that the method performs better when the loss is important. The approach produces satisfying results on both real and artificial data for the set of tested parameters.

Comparison of Two Swarm Intelligence Optimization Algorithms on the Textual Color Problem for Web Accessibility

Currently, web accessibility is not a major concern of webmasters while creating web sites. For disabled people, it rapidly becomes an obstacle to inclusion in the society. Identifying and circumventing existing barriers constitute an important research topic. In this work, we are concerned with the problem of color accessibility of textual contents in web pages. In many cases, the textual colors of a web page do not respect the minimum constraints defined by recommendations like WCAG 2.0. For example, WCAG 2.0 requires that a minimum difference of brightness, tonality and contrast is ensured. Using the Smart Web Accessibility Platform, we try to transform the colors using a client-side HTTP proxy the best possible while retaining a reasonable access time for the web content. To solve the textual color problem for accessibility, we adapt two swarm intelligence based optimization methods (ABC and API) and we hybridize them with a line search.

Annotation Tool for the Smart Web Accessibility Platform

Active and passive accessibility are two manner to improve web accessibility. While active accessibility mostly relies on norms and recommendations, it is practically proved that it is not sufficient. Passive accessibility is achieved by a posteriori content transformations. The Smart Web Accessibility Platform (SWAP) is a set of open source tools designed to tackle the passive accessibility problem of web contents. This article presents the goals and aims of SWAP through its main components: the proxy, the server and the annotation tool. The annotation tool is built using the proxy of SWAP. We explain how such design allows the annotation tool to be maintainable, independent of the browser and very flexible compared to other design.

Nature Inspires New Algorithms

Nature modeling is a leading trend in optimization methods. While genetic algorithms, ant-based methods, and particle swarm optimization are well-known examples, there is a continuous emergence of new algorithms inspired by nature. In this chapter, we give a short overview of the most recent promising new algorithms.