Essa semana o Prof. Don Towsley, nosso colaborador de mais de 15 anos, estará visitando o PESC através de um projeto de cooperação NSF-MCTIC (professores participantes da UFRJ: Rosa Leão, Daniel Sadoc Menasche e Edmundo).

O Prof. Towsley é um dos pesquisadores que mais publicam considerando TODA a área de Computação, com 64.547 citações de acordo com o Google Scholar (click here).

As suas áreas de intereesse são: networking, network science, performance evaluation

Mas a sua palestra (veja o abstract) tem interesse para o pessoal de IA e outras áreas.

=========================================================================

Titulo: Classical and Quantum Diffusion Convolutional Neural Networks

Palestrante: 

   Professor Don Towsley,

   College of Information & Computer Sciences

   University of Massachusetts

   Amherst, MA 

Dia: quinta feira desta semana: 18 de Outubro

Horário: 10-12.

Local: a princípio sala 324-B, no horário da aula de Aprendizado de Máquina.

ABSTRACT

In this talk we present diffusion-convolutional neural networks (DCNNs), a new model for graph-structured data. Through the introduction of a diffusion-convolution operation, we show how diffusion-based representations can be learned from graph-structured data and used as an effective basis for node classification. We focus first on DCNNs that rely on random walks for extracting and learning information. They exhibit several attractive qualities, including a latent representation for graphical data that is invariant under isomorphism, as well as polynomial-time prediction and learning that can be represented as tensor operations and efficiently implemented on a GPU. Through several experiments with real structured datasets, we demonstrate that DCNNs are able to outperform probabilistic relational models and kernel-on-graph methods at relational node classification tasks.  

We then focus on DCNNs that rely on quantum walks called quantum walk neural networks (QWNNs). A QWNN.  A QWNN learns a quantum walk on a graph to construct a diffusion operator which can then be applied to extract and learn information. We demonstrate the use of QWNNs on a variety of prediction tasks on graphs involving temperature, biological and molecular datasets and observe that it does at least as well and often better than its classical counterparts.

BIOGRAPHY

Don Towsley holds a B.A. in Physics (1971) and a Ph.D. in Computer Science (1975) from University of Texas.  He is currently a Distinguished Professor at the University of Massachusetts in the College of Information & Computer Sciences.  He has held visiting positions at numerous universities and research labs including INRIA. His research interests include networks, network science, and performance evaluation.

He was a founding Co-Editor-in-Chief of the new ACM Transactions on Modeling and Performance Evaluation of Computing Systems (ToMPECS) and has served as Editor-in-Chief of the IEEE/ACM Transactions on Networking, and on numerous editorial boards.  He has served as Program Co-chair of several conferences including INFOCOM 2009.  He is a member of ACM and IEEE.

He has received numerous IEEE and ACM awards including the 2007 IEEE Koji Kobayashi Award, and the ACM SIGCOMM and ACM SIGMETRICS Achievement Awards. Last, he has been elected Fellow of both the ACM and IEEE and is a corresponding member of the Brazilian Academy of Sciences.