You are here

Profile of a Scientist: Dr. Partha Pratim Talukdar: Teaching machines to learn by reading the web. A recipient of the Accenture Open Innovation Grant, 2015

When we read a book or watch a movie, our brain processes the sensory input and uses pre-acquired background knowledge to enrich our understanding of the content. Such background knowledge is normally not available for computers, which limits their ability for automated tasks such as translation of a document from one language to another. According to Dr. Partha Pratim Talukdar of Indian Institute of Science in Bangalore, such knowledge can be found from the ever-growing textual content on the Internet.

As an acknowledgement of his research efforts so far and his potential to go further, Accenture Technology Lab has recently awarded to Dr. Talukdar their Open Innovation Grant. This award includes financial assistance, as well as the opportunity to collaborate with scientists in Accenture who are working on related problems. This year, this award has been provided 11 grants to leading universities across the world, including 4 from India – IISc, IIT Madras, IIT Bombay and BITS Pilani.

“We, at Accenture Technology Lab are very excited to award the University Grant to IISc Bangalore. University grant is a key program in our open innovation agenda and I am certain the outcome of the joint research will further advance our thought leadership”, said Sanjay Podder, Managing Director and Head for Software Engineering Research in Accenture Technology Lab.

Dr. Talukdar’s main research interest are in Natural Language Processing, Machine Learning, and Cognitive Science. He holds a PhD from University of Pennsylvania, and has had Post-Doctoral research experience in Carnegie Mellon University. Last year, he joined IISc as a faculty member, and has established the Machine And Language Learning (MALL) Lab.

Dr. Talukdar’s research is aimed at developing algorithms by which computers can extract and exploit the textual content of the web according to the needs of various applications. This process has a large number of challenges, since the user-generated textual content lack uniformity and structure. They may also be written in a wide number of languages, and often do not conform to standard rules of grammar. Moreover, their ever-increasing volume poses computational challenges. Dr. Talukdar, in collaboration with other experts in the field of Natural Language Processing and Machine Learning, is trying to alleviate such problems. His research is particularly relevant in Indian context, where there are huge volumes of textual content in multiple languages.

A related research interest of Dr. Talukdar is to study the cognitive processes in the brain when it deals with language. Making use of brain imaging techniques such as fMRI, he is trying to understand how the brain searches its store of knowledge while reading a text document. Such techniques, he believes, can be borrowed or adapted in designing algorithms for computers. For this purpose, he has been collaborating actively with neuroscientists in IISc as well as outside.

For more information, visit: http://talukdar.net.

Contact: partha@talukdar.net