https://graphpower.inria.fr/

Context
We are surrounded by large-scale interconnected systems, from the Internet to the power grid and social networks. While essential, the management of such networked systems is exceedingly hard mainly because of their intrinsic and constantly growing complexity. To overcome this challenge, data-efficient online decision strategies under uncertainty for high-dimensional and dynamic networks need to be designed. This workshop will bring together researchers from UCL and France research centers/universities aimed at answering this need from a common perspective: applying the tenets of graph signal processing to online sequential decision strategies (and machine learning at large).

Aims
Understand the potential of graphs in ML algorithms: exploiting the structure of ML problems can be of great impact on the efficiency of the learning mechanisms. During the workshop we will investigate graph signal processing tools (e.g., spectral or sparse representation, graph denoising) applied to ML strategies (e.g., bandit problems, reinforcement learning, and neural networks).

Identify key open problems on graph-based ML: discussions on the topic will lead to identifying key challenges that are still unsolved in our community and yet vital for the development of efficient sequential strategies.

Identify key applications: the strength of the envisioned collaboration among the attendees is the background complementarity (expertise of the participants in both fundamental or applied ML research). Hence, it will be key to identify applicative scenarios for fundamental research on graph-based ML.

The power of graphs in machine learning and sequential decision-making