Deep Probabilistic Models (Wake/Sleep)

Add to your list(s) Download to your calendar using vCal

• Yan Wu; David Barrett
• Thursday 04 December 2014, 15:00-16:30
• Engineering Department, CBL Room 438.

If you have a question about this talk, please contact mv310.

Many of the recent breakthroughs in training deep neural networks have used supervised learning. However, these methods do not exploit the vast amount of unlabeled training data that is becoming available. A solution to this problem is to use unsupervised learning, but this is difficult in deep neural networks because inferring hidden variables in deep layers is usually intractable.

The wake-sleep algorithm is an unsupervised learning algorithm that learns to infer hidden variables using two modes of operation: a wake phase, in which a network is driven by training data, and a sleep phase in which the network generates fantasy data. We will describe the wake-sleep algorithm for two neural network architectures: The Helmholtz Machine and Deep Belief Nets. These are two of the most influential neural network architectures for unsupervised learning in neural networks.

This talk is part of the Machine Learning Reading Group @ CUED series.

This talk is included in these lists:

• All Talks (aka the CURE list)
• Cambridge Centre for Data-Driven Discovery (C2D3)
• Cambridge Forum of Science and Humanities
• Cambridge Language Sciences
• Cambridge University Engineering Department Talks
• Cambridge talks
• Centre for Smart Infrastructure & Construction
• Chris Davis' list
• Computational Continuum Mechanics Group Seminars
• Engineering Department, CBL Room 438
• Featured lists
• Guy Emerson's list
• Hanchen DaDaDash
• Inference Group Journal Clubs
• Inference Group Summary
• Information Engineering Division seminar list
• Interested Talks
• ML
• Machine Learning Reading Group
• Machine Learning Reading Group @ CUED
• Machine Learning Summary
• Quantum Matter Journal Club
• Required lists for MLG
• School of Technology
• Simon Baker's List
• TQS Journal Clubs
• Trust & Technology Initiative - interesting events
• bld31
• custom
• ndk22's list
• ob366-ai4er
• rp587
• yk373's list

Note that ex-directory lists are not shown.