Daniele Calandriello

I am a PhD student in the SequeL Team, under the co-supervision of Michal Valko and Alessandro Lazaric.

Research topic

My research focuses on efficient sequential learning in structured and constrained environments.
Sequential learning is efficient when the learning process can scale to large problem instances, while the constraints on the environment can be computational, such as limited memory to store the problem and time to find a solution, or statistical, such as limited amount of data to learn on. My goal is to develop new approximate algorithms that, using only a fraction of the resources, can compute an approximate solution that is provably close to the one of the original algorithm.

In particular, I focus on scalable optimization and linear algebra methods using incremental approximation and sketching. I am also interested in Reinforcement Learning and Learning on Graph.

More details in my CV.