Discrete Mathematics and Informatics

This research takes place as a collaboration between the Transport and Mobility Laboratory (TRANSP-OR) at EPFL and the industrial partner PSA Peugeot Citroen. The objective of the research project is to develop new methodologies to improve forecasting for market shares of different car makes, accounting for institutional, household and attitudinal factors. The collaboration between PSA and TRANSP-OR is designed to combine the research objectives with the short and medium term needs of PSA for improved market forecasting. The semester project is focused on the latent dimensions of the vehicle choice process. In fact, there are indications in the literature that planning, acquisition and use of a vehicle is influenced by features such as attitudes, life-changes, status aspirations and perceptions of making a good deal. Drawing on an existing large dataset containing rich information from new car buyers on opinions and attitudes I worked towards gaining a better understanding of the wider motivations for vehicle acquisitions.

Developments in Robotics and Computer Animation have shown that the research of Bipedal locomotion is of great importance and necessity in both fields. But it has proven to be a quite complicated task to achieve fluid human like locomotion. The approach presented in this paper is based on the empowerment of the standard control mechanism, generally used in Robotics, with a human like muscle control system and utilizes an Evolutionary technique to evolve a bipedal walker. As a result a number of controllers have been evolved, which are able to control the walker for varied distances. The resulting walkers have then been analysed for apparent differences, which are brought about by utilizing the above mentioned muscle system in the evolutionary process.

The global instruction scheduler and software pipeliner implemented by ISP RAS for the GCC compiler based on the selective scheduler has some deficiencies. Among some of the important ones we can name the absence of predication support for IA-64 and the usage of implicit data dependencies (i.e., no data dependence graph (DDG) is constructed and supported in the process of scheduling, but rather the elementary operation of moving an instruction up through another one is supported). This paper describes how we deal with both deficiencies, first, by adding predication support to the scheduler as yet another variant of an instruction transformation, which is a novel contribution of the paper, and second, by suggesting the explicit DDG construction and maintenance algorithm within the created scheduling framework in the GCC compiler, ∗Institute for System Programming of Russian Academy of Sciences, {dm, amonakov, abel, tigran, mamikon}@ispras.ru

The Job market is an ever moving and evolving entity. So are the competencies and qualifications it demands of prospective employees. In our previous work we modelled these competencies using topic models, but in order to have a more effective understanding of the market, we have to take into account the dynamics of the system as well. Here we propose using dynamic topic modelling as a means of analysing the job market for competencies and qualifications stemming from our previous research.

Engineering education is often presented as frameworks made up of intended competences based on product lifecycle step models. Such are the Conceiving-Designing-Implementing-Operating(CDIO) syllabus and the European e-Competence Framework (e-CF). In this article we analyze the correlation between the expert created competences implemented in these models and the job market requirement quali¯cations represented as topics extracted with probabilistic topic models.

Analysis of the job market quali¯cation requirements is crucial for several various institutions responsible for the job market regulation as well as educational institutions. In this work we introduce the concept of quali¯cation requirements extraction from a large set of job descriptions presented in open format based on probabilistic topic models. This model is then used to gather valuable insight about the current trends in the IT, Financial and Engineering industries.

The Job market is an ever moving and evolving entity. So are the competencies and qualifications it demands of prospective employees. In our previous work we modelled these competencies using topic models, but in order to have a more effective understanding of the market, we have to take into account the dynamics of the system as well. Here we propose using dynamic topic modelling as a means of analysing the job market for competencies and qualifications stemming from our previous research.