Research Interests


Low Power Circuit Design Techniques for Machine Learning Hardware

Janakiraman If you are looking to join me for a PhD, you'll most likely work in this field unless you have a more exciting problem at hand that is of mutual interest. I am particularly interested in looking at fundamental circuit level building blocks for machine learning right from feature extraction to classification. Initially the group will focus on hardware implementation of trained neural networks and later move to training in hardware. This will involve a good understanding of machine learning algorithms and low power circuit design techniques. The focus of my group will be to build a complete system eventually. A combination of courses from Computer Science and Electrical Engineering departments at IITM will help you tackle these problems with great confidence

Statistical Analysis in VLSI

Janakiraman I am interested in select Design Automation problems. In particular, problems related to fast monte carlo, Statistical Leakage and Timing analysis. Solution to these problems require firm basics in probability theory and good amount of mathematical analysis. That said, making the right approximations based on the VLSI problem at hand is equally important and hence you need to have a good understanding of VLSI as well. The EE department offers excellent courses to bring you up to speed in these areas.

Projects


Solenoid Control for a Gasoline Pump Injector (Jan - Jul'17)

Company: Stanadyne India Pvt. Ltd
Project Associate: Gourab Ghosh and Ashwin Sundar