RAPDARTS: Resource-Aware Progressive Differentiable Architecture Search

no code implementations • 8 Nov 2019 • Sam Green, Craig M. Vineyard, Ryan Helinski, Çetin Kaya Koç

Gradient-based approaches impose more structure on the search, compared to alternative NAS methods, enabling faster search phase optimization.

Neural Architecture Search

Distillation Strategies for Proximal Policy Optimization

no code implementations • 23 Jan 2019 • Sam Green, Craig M. Vineyard, Çetin Kaya Koç

DQN distillation extended the original distillation idea to transfer information stored in a high performance, high capacity teacher Q-function trained via the Deep Q-Learning (DQN) algorithm.

Q-Learning Reinforcement Learning (RL)

Visual Diagnostics for Deep Reinforcement Learning Policy Development

no code implementations • 14 Sep 2018 • Jieliang Luo, Sam Green, Peter Feghali, George Legrady, Çetin Kaya Koç

In this paper, we present our extensions of CNN visualization algorithms to the domain of vision-based reinforcement learning.

reinforcement-learning Reinforcement Learning (RL)