Feedback

Ashwin Srinivasan

Senior Professor

AI for Science, Neuro-Symbolic Learning, Symbolic Machine Learning
Rm D-167, Department of Computer Science and Information Systems
BITS Pilani K.K. BIrla Goa Campus
National Highway 17B, Zuarinagar
Goa 403726

Research Miscellany

Some research grants I have been involved with:

  • 2023: PI on an Industrial Research Agreement as TCS Affiliate Professor for INR 3,900,000; co-PI on a New Venture Fund grant for USD 350,000 to develop a training program for undergraduates and working professionals in Data Science in Climate and Health
  • 2022: Co-PI on Department of Biotechnology grant ‘DBT project BT/PR40236/BTIS/137/51/2022 ”Developing Predictive Models for ’druglikeness’ of small molecules” for approximately INR 9,600,000
  • 2021: PI on an Industrial Research Agreement as TCS Affiliate Professor for INR 3,120,000
  • 2020: PI on an Industrial Research Project with Reflexis Corp. for approximately INR 2,160,000; PI on an Industrial Research Grant with TCS (DataLab3) for approximately INR 1,275,000; PI on a TCS Innovation Grant for approximately INR 1,500,00)
  • 2019: I was the main technical lead for the University in a successful grant proposal for Alumni funding of USD 1,00,000 for a Centre; PI on TCS Industrial Projects (DataLabs1,2) for approximately INR 5,000,000.
  • 2018: Erasmus+ mobility grant for exchange of faculty and students between Goa and Porto (approximately 20,000 EUR).
  • 2017: I am the only investigator in a CORE research grant EMR/2016/002766 on “Knowledge-Rich Deep Models for Optimisation”. This is for approximately 2,300,000 awarded by the Indian Science and Engineering Research Board (SERB); TCS Research Advisory for approximately INR750,000.

My current research interests:

  • AI for Scienceincluding the use of ML models for computational drug-discovery; and constructing logical models of systems from their behaviour using meta-interpretative learning;
  • Neural-Symbolic Learning: Constructing models that combine deep neural networks with symbolic learning. This examines both the inclusion of domain-knowledge into neural-networks through the use of symbolic learning; and the construction of explainable models for deep networks;
  • Intelligible AI: formal models for 2-way intelligibility between agents that predict and explain. This has applications to the design of human-in-the-loop AI systems, which require both human and machine understand the information they provide to the other.