[AIRG] Resource Constrained Neural Network Architecture Search

[Yunyang Xiong <yxiong43@xxxxxxxx>]

Hello,

I will be presenting our new work, resource constrained neural network architecture search, on this Wednesday. 

Title: resource constrained neural network architecture search (will be appear on arXiv soon)

Authors: Yunyang Xiong, Ronak Mehta and Vikas Singh

Presenter: Yunyang Xiong

Time: 4 pm, Wednesday, 12/5, 2018

Location: CS 3310

 

The design of neural network architectures is frequently either based on human expertise using trial/error and empirical

feedback or tackled via large scale reinforcement learning strategies. Most methods in use require exorbitant computational resources. 

The above challenges are exacerbated because the search procedure must now balance accuracy with certain budget constraints on resources. 

Our goal in this work is to  formulate resource constrained architecture search as a set function optimization, and design a heuristic based on ideas known 

to be effective for constrained submodular optimization.

Best regards,

Yunyang