[Adapt] Abstract of the talk in this seminar

赵凯祺 kaiqi_zhao at 163.com
Wed Apr 17 10:25:52 CST 2013


Hi adapters,


In this seminar, I will give a talk about learning deep belief net. Here's the abstract:


In this big data era, labeled data is too sparse compare to the input data we face. Traditional machine learning approaches and & models relies heavily on the labels. Neural networks models the structure of human brains. Researchers tried to learn deep neural networks but failed in the past because when the layers increase, the label data they need increase exponentially. In recent years, belief nets are widely used in neural network because the learning process is unsupervised. And a great discovery by Hinton, etc. that infinit sigmoid belief net is a restricted Boltzmann Machine makes the learning of deep belief net very easy. I am going to introduce belief network and deep learning algorithms on it.




Cheers!
Kaiqi
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://cs.sjtu.edu.cn/pipermail/adapt/attachments/20130417/61b49c6d/attachment.html>


More information about the Adapt mailing list