*********************************
There is now a CONTENT FREEZE for Mercury while we switch to a new platform. It began on Friday, March 10 at 6pm and will end on Wednesday, March 15 at noon. No new content can be created during this time, but all material in the system as of the beginning of the freeze will be migrated to the new platform, including users and groups. Functionally the new site is identical to the old one. webteam@gatech.edu
*********************************
Title: Disentangling Neural Networks Representations for Improved Generalization
Michael Cogswell
Ph.D. Student
School of Interactive Computing
Georgia Institute of Technology
Date: Thursday, August 29th, 2019
Time: 3:00pm - 5:00pm (EST)
Location: CODA C0915 Atlantic
Committee:
Prof. Dhruv Batra, School of Interactive Computing, Georgia Institute of Technology
Prof. Devi Parikh, School of Interactive Computing, Georgia Institute of Technology
Prof. James Hays, School of Interactive Computing, Georgia Institute of Technology
Prof. Ashok Goel, School of Interactive Computing, Georgia Institute of Technology
Prof. Stefan Lee, Oregon State University
Abstract:
Modern neural networks have helped the field of artificial intelligence tackle increasingly complex perceptual problems. However, despite the increasingly broad capabilities of neural networks, each new task they are applied to still requires significant engineering effort. This is in part due to the entangled nature of the representations learned by these models. Entangled representations capture spurious patterns that are only useful for specific examples instead of factors of variation that explain the data generally. We show that encouraging representations to be disentangled makes them generalize better.
In this thesis proposal we identify three notions of entangled representations, enforce disentanglement in each case, and show that more general representations result from enforcing disentanglement. Our existing work considers language compositionality in goal driven dialog and statistical independence of features in image classification as notions of disentanglement. Our proposed work will disentangle a multi-modal language and vision representation from the tasks it is used to solve. By increasing the generality of neural networks through disentanglement we hope to reduce the effort required to apply neural networks to new tasks.
