Stony Brook University Computer Science

Prof. Byungkon Kang will be giving a talk on "Mutual Equilibrium Models".

Please find the seminar details below:
Title: Mutual Equilibrium Models
Date & Time: Friday, October 23, 2 PM
Venue: B206

Abstract
This work proposes a mutual feedback architecture, MEQ, that refines the two inputs, of possibly different modalities, into a pair of coupled embeddings such that each embedding reflects the information of the other.
The core idea is to incorporate continuous interchange of information between the two inputs. This leads to a mutual feedback architecture consisting of two components whose outputs are fed back into the other.
The final output of this model is defined as the fixed point of this interaction. Theoretical analysis of this model offers interpretation as well as design choices to prevent failure cases.
The benefits of MEQ is shown through classification and visual grounding tasks spanning various datasets.