Home > Seminars > Taylor Berg-Kirkpatrick - Unsupervised Models for Language, Music, and Shakespeare

Taylor Berg-Kirkpatrick - Unsupervised Models for Language, Music, and Shakespeare

Start:

12/7/2017 at 3:30PM

End:

12/7/2017 at 4:45PM

Location:

140 DeBartolo

Host:

College of Engineering close button
headerbottom

David Chiang

David Chiang

VIEW FULL PROFILE Email: dchiang@nd.edu
Phone: 574-631-9441
Website: http://www.nd.edu/~dchiang/
Office: 326D Cushing Hall

Affiliations

Department of Computer Science and Engineering Associate Professor
College of Engineering Associate Professor
Natural language processing, machine learning, and digital humanities.
Click for more information about David
David Chiang
dchiang@nd.edu
574-631-9441
Add to calendar:
iCal vCal
Without careful consideration of the relationship between input and output, unsupervised learning problems can be under-constrained. This talk will discuss ways to make unsupervised problems feasible by incorporating different types of inductive bias. First, we focus on a set of tasks related to the digital humanities, including historical document recognition, music transcription, and compositor attribution. For each of these tasks, strong prior knowledge about the causal process behind the data can be encoded into the model. We show how to leverage casual knowledge as a source of constraint to make these learning problems feasible. Finally, we investigate tasks where causal structure is difficult to encode. Here, we instead rely on neural parameterizations to define data-generating distributions and discuss the corresponding challenges. 

Seminar Speaker:

Taylor Berg-Kirkpatrick

Taylor Berg-Kirkpatrick

Carnegie Mellon University

Taylor Berg-Kirkpatrick joined the Language Technologies Institute at Carnegie Mellon University as an Assistant Professor in Fall 2016.  Previously, he was a Research Scientist at Semantic Machines Inc. and, before that, completed his Ph.D. in computer science at the University of California, Berkeley.  Taylor's research focuses on using machine learning to understand structured human data, including language but also sources like music, document images, and other complex artifacts.