Compositional Models - Complexity of Representation and Interference

Explore the intricacies of compositional models in this 51-minute lecture by Alan Yuille from the University of California, Los Angeles. Delve into the complexity of representation and interference, covering topics such as data-driven models, feedforward and feedback mechanisms, generative models, and hierarchies. Examine the history of generative models and their application to cluttered data scenarios. Analyze cross-examples, parameters, and local ambiguity in various model types. Gain insights into the importance of hierarchical structures in compositional modeling through in-depth discussions and examples.

Introduction
Datadriven models
Representations
Feedforward Feedback
Do generative models require feedback
History of generative models
Models of data in clutter
Hierarchies
Cross
Examples
Parameters
Other Models
Local Ambiguity
Why Hierarchies
Discussion