We are interested in making smart machines that can be used reliably by humans in everyday life. Artificial Intelligence (AI) has made tremendous progress since the days of expert systems and has advanced the research in several areas - cognitive science, natural language processing, game theory, machine learning, speech recognition, activity recognition and decision-theory to mention a few. Historically, AI has used either the logical approach (to address structured problems) or the statistical approach (to handle uncertainty). Recent years have witnessed a tremendous development of techniques to handle large-scale, structured and uncertain domains. Our research interests lie in the advancement and application of algorithms from the exciting combination of logical and statistical AI in several related areas and can be divided into three intersecting themes:
Efficient Statistical Relational Learning:
Traditional machine learning approaches assume that examples are generated uniformly and independently from a large pool of data. This assumption may not be applicable in several real-world problems. Instead, it is common to observe rich relational structure in data. For example, in the medical domain, a strong relationship exists between doctors, patients, symptoms, diagnoses, and prescribed drugs. We consider the problem of learning in the presence of rich, multi-relational, semi-structured data.
Decision Theoretic Learning:
The goal of Reinforcement Learning (RL) is to build learning agents that are connected to their environments through perception and action. We use RL in two different kinds of problems – one in which the agent explores the environment and learns to act and the second in which an expert provides training to the agent and the agent learns to mimic the expert.
Application of the above ideas for bio-medicine:
In particular, we are interested in predicting heart attacks of patients in adulthood given their attributes in their early youth. We are also working on developing smart communicative devices for disabled people such as people with dementia, Alzheimer’s, autism etc.