PhD Defense: "PhD Defense: “Scalable Front End Designs for Communication and Learning”"

Aseem Wadhwa

December 4th (Thursday), 2:00pm
Harold Frank Hall (HFH), Room 4164 (ECE Conf. Rm.)

In this talk, we consider three classical estimation and detection problems that face new scalability challenges. We discuss front end designs which are customized to the specific application at hand and make the systems more scalable and efficient. The first two case studies pertain to the canonical problems of synchronization and equalization for communication links. As the system bandwidths scale, challenges arise due to the limiting resolution of analog-to-digital converters (ADCs). We discuss new architectures that react to this bottleneck by drastically relaxing the precision requirements of the front end and appropriately adapting the back end algorithms using Bayesian principles. We demonstrate the importance of using a dither prior to quantization and carefully designing the quantization thresholds. The third problem we discuss belongs to the field of computer vision. Inspired by the research in neuroscience about the mammalian visual system, we redesign the front end of a machine vision system to be neuro-mimetic, followed by layers of unsupervised learning using k-means clustering. This results in a framework that is intuitive, easier to implement compared to the approach of supervised deep networks, and amenable to the increasing availability of large amounts of unlabeled data. Supervised classification, using a generic support vector machine (SVM), is applied at the end. We obtain competitive classification results on standard image databases NORB and MNIST.

About Aseem Wadhwa:

Aseem Wadhwa received his Bachelors degree in Electrical Engineering from the Indian Institute of Technology Delhi in 2009 and Masters in Electrical and Computer Engineering from UCSB in 2011. He joined Prof. Madhow's research group and has worked on various projects ranging from channel modeling for UAV routing, communication system design, signal processing for FMRI, to more recently computer vision. He interned with the modem group at Qualcomm Santa Clara in 2012. His research interests include communication systems, signal processing and machine learning.

Hosted by: Professor Upamanyu Madhow, WCSL Lab