| Title | : | Stochastic computation using Neural Networks |
| Speaker | : | Sarath Chandar (University of Montreal, Canada) |
| Details | : | Thu, 23 Mar, 2017 2:00 PM @ CS25 |
| Abstract: | : | Abstract:
Deep Neural Networks (DNNs) are powerful parametric models that have
significantly improved the state-of-the-art in object recognition,
speech recognition, machine translation, image generation, etc. In
this tutorial, I will introduce stochastic neural networks which can
contain both deterministic functions and conditional probability
distributions. Stochastic neural nets combine the power of neural
networks with graphical models and are more expressive. However,
backpropagation with stochastic units is not trivial due to their
non-differentiable nature. We will discuss a generic framework for
backpropagation in stochastic computation graphs and show how various
recently proposed methods for training stochastic neural networks are
special case of this generic procedure. We will also discuss various
case studies in attention, memory, and control.
Target Audience: anyone with basic knowledge in neural networks (backprop, feedforward/recurrent/conv nets). Basic graphical models knowledge is desirable but not required to understand the talk. Speaker Bio: Sarath Chandar is currently a PhD student in University of Montreal under the supervision of Yoshua Bengio and Hugo Larochelle. His work mainly focuses on Deep Learning for complex NLP tasks like question answering and dialog systems. He also investigates scalable training procedure and memory access mechanisms for memory network architectures. In the past, he has worked on multilingual representation learning and transfer learning across multiple languages. His research interests include Machine Learning, Natural Language Processing, Deep Learning, and Reinforcement Learning. Before joining University of Montreal, he was a Blue Scholar in IBM Research India for a year. He has completed his MS by Research in IIT Madras. To view the complete publication list and speaker profile, please visit: http://sarathchandar.in/ |
