Seminar: Parallel and Distributed Inference in Coupled Tensor Factorization Models

You are cordially invited to the following seminar in the Computer Engineering Department.

Speaker: Doç. Dr. S. Ilker Birbil
Faculty of Engineering and Natural Sciences,
Sabanci University
Web: http://people.sabanciuniv.edu/sibirbil

Date: February 27th, 2015 (Friday)
Time: 13:45
Place: Computer Engineering Department, Seminar Room

Abstract:
Coupled tensor factorization methods are useful for sensor fusion, combining in- formation from several related datasets by simultaneously approximating them by products of latent tensors. These methods are naturally adopted for the mod- ern computing infrastructure that comprises of systems with hybrid architectures where the data might be very large and be distributed across several computers where each computer has multiple processors such as a multicore system or a graphics processing unit (GPU). In this paper, we present a parallel and distributed algorithmic framework for coupled tensor factorization to simultaneously estimate latent factors, specific divergences for each dataset as well as the relative weights in an overall additive cost function. We illustrate the proposed method on syn- thetic data experiments, where we show that the speed of the proposed algorithm is nearly proportional to the number of processing nodes.

This is a joint work with Ali Taylan Cemgil, Umut Şimşekli, Beyza Ermiş (Boğaziçi University) and Figen Öztoprak (Bilgi University)

Kind regards,

Yrd. Doç. Dr. Mustafa Ozuysal
Bilgisayar Mühendisliği Bölümü