Categoria: Seminari e Convegni
Stato: Corrente
5 maggio, ore 14.00

Toward (Truly) Resilient Neural Networks in High-Stakes Mobile Systems

Sala Riunione 4, quarto piano del Dauin

Seminario dal titolo Toward (Truly) Resilient Neural Networks in High-Stakes Mobile Systems, tenuto da Francesco Restuccia della Northeastern University (USA).

Quando: Lunedì 5 maggio, ore 14:00
Dove: Sala Riunione 4, quarto piano del Dauin

Tema dell’incontro: Mobile autonomous systems extensively leverage neural networks to implement a wide variety of high-stakes inference and control tasks (e.g, obstacle detection for unmanned navigation). As such, guaranteeing that neural networks will be resilient to both intentional and unintentional perturbations becomes necessary. Most of existing approaches tackle this issue at training time, which cannot address sudden and/or unpredictable perturbations. While test-time adaptation has been proposed, existing work takes a series of unrealistic assumptions that does not make them applicable in mobile autonomous systems. In this seminar, we are going to present our recent research results on ensuring resilience in neural networks operating in the context of mobile systems. We will conclude the seminar with discussions on ongoing research efforts and possible research directions.

Bio: Francesco Restuccia is an Assistant Professor in the Department of Electrical and Computer Engineering at Northeastern University. Prof. Restuccia’s main research focus is addressing the fundamental challenges related to edge-assisted data-driven resilient mobile systems. Prof. Restuccia’s research is funded by several grants from the National Science Foundation and the Department of Defense. Prof. Restuccia has received the ONR Young Investigator Award, the AFOSR Young Investigator Award, the ACM SIGMOBILE Research Highlights Award, the Mario Gerla Award in Computer Science, as well as best paper awards at IEEE INFOCOM (twice) and IEEE WOWMOM. Prof. Restuccia has been granted 12 US patents and has been cited 4700+ times with an h-index of 36. He regularly serves as a TPC member and reviewer for several top-tier ACM and IEEE conferences and is in the editorial board of Computer Networks, IEEE Transactions on Cognitive Communications and Networking and IEEE Transactions on Mobile Computing. He is a Senior Member of IEEE and ACM.