IEEE BUET Student Branch  SPS

Technical Talk On Convolutional Coding For Distributed Matrix Multiplication

who we are

At a Glance


July 29, 2019


Room 632, ECE Building, BUET

Number of Participants


Guest Speaker

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

IEEE SPS BUET SB Chapter, IEEE BUET Student Branch, IEEE BDS and IEEE SPS BD Chapter arranged a technical talk on July 29, 2019 on convolutional coding for distributed matrix multiplication by Anindya Bijoy Das, PhD student, Electrical and Computer Engineering, Iowa State University. The event was arranged as an additional segment of the orientation of IEEE members and executive committee of IEEE BUET SB, WIE AG and technical chapters (IAS / PES / EDS / RAS / SPS / CS) for 2019-20. The speaker is currently researching in the area of the topic with his PhD supervisor Professor Dr. Aditya Ramamoorthy. Professor Dr. Celia Shahnaz, Chairperson of IEEE BDS and adviser of IEEE SPS BUET SB Chapter, was present during the talk. 


Upon the request of Professor Celia Shahnaz, Anindya Bijoy Das started his talk by sharing about his academic life in higher studies abroad. Mr. Anindya shared how Iowa and his city Ames is among the cities of lowest crime rate, being mostly resided by students attending university such as Iowa State University. Besides, he encouraged the attendees to participate in conferences abroad during student life to improve the research culture. He also talked about his department and about the professors researching in areas involving signal processing and machine learning. His current research interest involves straggler mitigation in distributed system, which deals with lagging or low performing computing units in a distributed computation system by allowing the rest of the computation to be retrieved from other units/workers. He explained how such works of mathematics and information theory are currently important in fields of signal processing and machine learning.


Since many of the participants were freshly admitted undergraduate students, Anindya Bijoy Das explained their method of multiplying large matrices in separate small workers / computers / threads in a comprehensive manner. He showed visualizations of multiplication of a matrix with blocks of rows in a column vector separately in different workers to generate the resulting matrix without requiring one large worker, improving time and memory constraints. 


Before starting with explanations of rigorous algorithms and equations, we were honored with a short note from Professor Mehdi Anwar, Electrical and Computer Engineering, University of Connecticut. He encouraged the students about higher studies and shared prospects. He also explained one of his recent works. Finally, he welcomed the students to attend these talks and trainings. 


At this point, Anindya Bijoy Sas started explaining their novel methods of matrix multiplication in distributed fashion while dealing with stragglers. By improving the computation approach of multiplication of large matrices such as it is required in machine learning algorithms, this method would help optimize time and memory resources properly in or among devices. Professor Celia Shahnaz also asked a question about applications of this technique, and Anindya Bijoy Das explained how this would help machine learning in signal processing in terms familiar to newcomers. Finally, Anindya Bijoy Das concluded the program by welcoming the participants to reach out to him in case of any queries regarding this research.