HC-Search: A Learning Framework for Search-based Structured Prediction
Monday, February 10, 2014 - 4:00pm - 4:50pm
KEC 1001
Janardhan Rao (Jana) Doppa
PhD. Student
School of EECS
Oregon State University
Abstract:
We are witnessing the rise of the ÂBig Data paradigm, in which massive
amounts of data (e.g., text, images, videos, speech) -- much of it collected as
a side-effect of ordinary human activity -- can be analyzed to make sense of
the data, and to make useful predictions. To fully realize the promise of Big
Data, we need automated systems that can transform structured inputs to
structured outputs (e.g., parsing a sentence, resolving coreferences of entity
and event mentions in a piece of text, interpreting a visual scene, translating
from one language to another). Problems such as these are often referred to as
structured prediction problems in the machine learning community. These
prediction problems pose severe learning challenges due to the huge number of
possible outputs (e.g., many possible parse trees for a sentence). In this
talk, I will introduce a new framework to solve these structured prediction
problems called HC-Search. The problem of structured prediction is fo!
rmulated as an explicit search process in the combinatorial space of outputs.
The search seeks to optimize the cost function C using a heuristic H to guide
the search. Both the cost function and the heuristic are learned from
supervised data to minimize a given task loss function. I show that my
HC-Search framework achieves state-of-the-art results in a wide range of
structured prediction problems that arise in natural language processing and
computer vision, exceeding the previous best results by significant margins. I
will close with some on-going work on applications of this framework and
challenging open problems.
Speaker Biography:
Janardhan Rao (Jana) Doppa is a final year PhD student with the Artificial Intelligence group at Oregon State University. He received his M.Tech degree in computer science from Indian Institute of Technology (IIT), Kanpur, India. His general research interests are in Artificial Intelligence (AI) and Machine learning. His dissertation explores how to integrate two fundamental branches of AI, namely learning and search to solve structured prediction problems arising in natural language processing (NLP) and computer vision (CV). He received an Outstanding Paper Award at the AAAI 2013 conference for his structured prediction work, and an Outstanding Graduate Research Assistant Award (2013) from the College of Engineering, Oregon State University.
_______________________________________________
Colloquium mailing list
[email protected]
https://secure.engr.oregonstate.edu/mailman/listinfo/colloquium
