Research Guide: Dr. B. Ravindran
Areas of Interest:
- Machine Learning/Data Mining
- Graph Mining
- Time series Analysis
Research: I am currently working on a project to study how patterns ("episodes") in different time series affect each other. An episode is defined as an ordered tuple of events. The main motivation of this study is to find whether episodes in a certain time series cause/affect episodes occuring in other time series. When such pairs of episodes are found, they are called a Coupled Episode Pair(CEP). CEPs that occur with high frequencies can be used to predict the outcome on the subsequent time series. Some challenges in this project include efficiently counting episodes and coupled episodes, using CEPs to predict future labels, etc.
I have also worked on a Sentiment Analysis project which employed collective classification techniques on a corpus of automobile reviews to gain fine-grained polarity information on various features of automobiles.