The LTS2 lab at Ecole Polytechnique Federale de Lausanne (EPFL) has an opening for a PhD student in the context of the SNSF project «Deep Learning for Graph-Structured Data».
- TOPIC. The project focuses on deep learning algorithms that utilize the graph-structure in data in order to recognize patterns and make relevant predictions. From the organization of human interaction (social networks), human interest (recommendation networks), human knowledge (Wikipedia), and human mobility (transportation networks), an astounding amount of data exist that posses graph structure. Within the machine learning community there has been some recent progress in building neural networks for graph-structured data, with the main thrust centered around generalizations of convolution. Continuing in this direction, the aim of this project is to step away from common intuitions and methodologies that deep learning relies on and rethink the fundamentals from a graph perspective.
- REQUIREMENTS. We are interested in students with EE, CS or mathematics background. Successful applicants need to be ambitious and driven, with a solid understanding of computer science (algorithms, graph theory) and a good grasp of fundamentals (linear algebra, probability, optimization). Good written and spoken English are essential. Advanced coding skills (Python) and former experience with deep learning and (graph) data analysis are highly desirable.
- BENEFITS. LTS2 provides a fun working environment. The lab combines expertise in data analysis, signal processing and machine learning. The position comes with a very competitive salary, contribution to pension plan, accident insurance and a range of further benefits. The working language is English.
Interested candidates should apply to the EDIC or EDEE doctoral programs of EPFL and list Prof. Pierre Vandergheynst as a possible PhD host. Applications via email or postal services will not be considered.
Starting date: as soon as possible
For more details please check: https://phd.epfl.ch/application
For inquiries, please contact Andreas Loukas at email@example.com