Friday, March 9, 2018

3:15 pm

Jordan Hall room 050

Sponsored by:
Department of Psychology

Speaker:  Damian Mrowca & Aran Nayebi, PhD students with Assistant Professor Dan Yamins, Department of Psychology, Stanford University

Aran Nayebi     

Title: Convolutional recurrent neural network models of dynamics in higher visual cortex

Abstract:Neurons in the ventral visual pathway exhibit behaviorally relevant temporal dynamics during image viewing. However, the most accurate existing computational models of this system are feedforward hierarchical convolutional neural networks (HCNNs), which capture neurons’ time-averaged responses, but do not account well for their complex temporal trajectories. Here we show that HCNNs augmented with both local and global recurrent connections are quantitatively accurate models of dynamics in higher visual cortex. In particular, we will compare a direct neural fit to the dynamics to a more normative task-optimized approach, and discuss subtleties with generalization, improved categorization performance, and choice of behavioral decoder.

Damian Mrowca

Title: Emergence of Structured Behavior from Curiosity-based Intrinsic Motivation

Abstract:Infants are experts at playing, with an amazing ability to generate novel structured behaviors in unstructured environments that lack clear extrinsic reward signals. In our work, we seek to replicate some of these abilities with a neural network that implements curiosity-driven intrinsic motivation. Using a simple but ecologically naturalistic simulated environment in which the agent can move and interact with objects it sees, the agent learns a world model predicting the dynamic consequences of its actions. Simultaneously, the agent learns to take actions that adversarially challenge the developing world model, pushing the agent to explore novel and informative interactions with its environment. We demonstrate that this policy leads to the self-supervised emergence of a spectrum of complex behaviors, including ego motion prediction, object attention, and object gathering. Moreover, the world model that the agent learns supports improved performance on object dynamics prediction and localization tasks. While our model has not yet been evaluated on real infants, our results are a proof-of-principle that computational models of intrinsic motivation might account for key features of developmental visuomotor learning in infants.

Friday, March 9, 2018
3:15 pm – 4:30 pm
Jordan Hall room 050

Humanities Seminar 

Faculty/Staff, Students, Alumni/Friends, Members