AnindyaTechTalk

A technical panel talk is being jointly organized by IEEE Signal Processing Society(SPS) Bangladesh Chapter and IEEE SPS BUET Student Branch as per the followings:

Date and Time: Monday, July 29, 2019; 04:00 PM
Venue:  , ECE Building, Dept. of EEE, BUET

All IEEE Members, IEEE SPS Members, Undergraduate and Post-graduate students are cordially invited to attend the talk.

Speaker:

Anindya Bijoy Das
PhD student
Electrical and Computer Engineering
Iowa State University
Ames, Iowa, USA

Speaker Biography:

Anindya Bijoy Das is a graduate student in Electrical and Computer Engineering department of Iowa State University, Ames, IA, USA since Fall 2016. He received his B.Sc. degree in Electrical and Electronic Engineering from Bangladesh University of Engineering and Technology (BUET), Dhaka, Bangladesh in 2014, and M. Engg. degree in Electrical Engineering from Iowa State University in 2018. He is currently working on straggler mitigation in distributed systems. His research interests include information theory and coding theory.

Topic:

Convolutional Coding for Distributed Matrix Computation

Abstract:

Distributed matrix computations (matrix-vector and matrix-matrix multiplications) are at the heart of several tasks within the machine learning pipeline. However, distributed clusters are well-recognized to suffer from the problem of slow or failed nodes. Prior work in this area has presented strategies based on polynomial interpolation to solve this problem, but such approaches, in practice, suffer from numerical problems.In our work, we present a strategy for distributed matrix computation based on convolutional coding. Our scheme can be decoded using a low complexity peeling decoder, and thus, the recovery process enjoys excellent numerical stability.