A MARIE SKŁODOWSKA-CURIE ACTIONS (MSCA) Innovative Training Network (ITN), TraDe-OPT offers 15 PhD positions (ESRs) within an innovative training program giving a solid mathematical background in (convex) optimization and data driven modeling combined with employability skills: management, fund rising, communication, and carrier planning skills. The trainign will be based on three modules: learning by research, learmning by courses, learning by doing. Embedded in the TraDe-OPT's training environment, each ESR will be developing an individual research project comprising significant advances in designing efficient algorithmic solutions and a related industry backed project.
The main scientific objective of the TraDe-OPT is to derive and analyse efficient optimization algorithms for solving data-driven problems. Applications to a broad range of social, economic, health and urban problems are expected. Nowadays, data production explodes: data are produced by a variety of sensors in industry, vehicles, scanners, internet and mobile devices. One of the emerging challenges is to extract interpretable information from these data. Currently, optimization, and especially, convex optimization, is at the core of many theoretical and algorithmic methods underpinning solver technologies for a myriad of data driven problems.
In this project we focus on the explicit modeling of relevant structures present in large data structures, preserving specificities such as the dynamical character of interactions, the multi-scale nature of problems, spatial information or the stochasticity of the observed process. The construction of mathematically sound models featuring one or several of these aspects will allow the extraction of rich information and underlying knowledge, with the help of dedicated machine learning techniques. Optimization will play a central role in every stage of this project, as the main computational technique. We will focus, in particular, on algorithms whose efficiency can be guaranteed, based on the exploitation of the above-mentioned structure present in our models, as well as additional characteristics such as sparsity, stochasticity or low-rank in the input data, or the use of different models of computations including parallel and distributed computing, with or without synchronization. Ideally, the algorithmic solutions developed within TraDE-Opt will scale to large, high dimensional datasets, will be stable with respect to perturbations of the data, and will satisfy the constraints imposed by privacy and security issues, while working with limited computational resources. The research program of TraDE-Opt is organized in three theoretical workpackages: WP1, WP2, WP3, and one applied workpackage WP4.
The proposed academic supervisors have already experience in supervising PhD students of different nationalities. They are highly qualified researchers (most of them have h-index >20 and more than 1000 citations) with related but different, research interests. They are all recognized at an international level, and actively participate to the dissemination activities on their research topics.For more informations about the team see here