Carnegie Mellon University
Swabha Swayamdipta is a postdoctoral investigator at the Allen Institute of Artificial Intelligence. Her research interests include representation learning for natural language processing and linguistic structure prediction. She received her PhD in Language and Information Technologies from Carnegie Mellon University in May 2019, advised by Noah Smith and Chris Dyer. During part of her PhD, she was a visiting graduate student at University of Washington in Seattle.
Prior to joining her PhD program, she earned a Masters degree from Columbia University. She received her Bachelors at the National Institute of Technology, Calicut, India. She has done research internships at Google AI, New York in 2017 and at Allen Institute of Artificial Intelligence in Seattle in 2018.
As the availability of data for language learning grows, the role of linguistic structure as well as patterns in our datasets need to be scrutinized. My research addresses each of these challenges. First, I have worked on a novel paradigm called scaffolded learning which enables us to leverage inductive biases from one structural source for prediction of a different, but related structure, using only as much supervision as is necessary. The resulting representations achieve improved performance across a range of tasks, indicating that linguistic structure remains beneficial even with powerful deep learning architectures. Moreover, even as NLP models in large data regimes report excellent performance, sometimes claimed to beat humans, we show that predictions are not a result of complex reasoning can be largely attributed to exploitation of some artifacts of data annotation, bringing about important discussions and challenges for the future.