Ongoing and future directions

Self-driving and autonomous vehicles

Autonomous vehicles have been a global obsession for the better part of the century. This technology is here to transform life on the road as we know it, save for a niggling question…how safe is it really? Car companies with automated vehicles are test driving for millions of miles. It would take hundreds of years to test existing fleets for fatalities and injuries prior to consumer use. How then, can we ensure that we can use these vehicles in our lifetimes?

In our research, we examine the entire picture of auto reliability and validation and color the  problems we think we need to care about in the 5-10 year time horizon. Different technologies in verification and reliability from our research apply to these problems.

Neural Networks

The panacea of deep learning has a problem- it is not interpretable. We are applying our insights from feature engineering and model analysis to understand what they generate. We are also evaluating these architectures for how reliable and secure they are to adversarial attacks.

GoldMine on AI/ML steroids!

We invented GoldMine in 2009.  This was much before the machine learning/AI revolution that has taken over the world in the past few years. While it’s produced great results with the machine learning algorithms we developed internally, we want to leverage the latest and greatest ML/AI algorithms to make GoldMine even better. We believe that the dose of AI/ML will make GoldMine generate system properties that are powerful and much higher quality than what a human being can.

 

Health-care analytics

After transforming the way we communicate, socialize, get entertained and shop, computing’s next target is healthcare. We are working on different aspects of health care- EMR, emotional health and physical health monitoring. We collaborate with hospitals and clinicians on data analysis, summarization, interpretation and validation.