/* ---- Google Analytics Code Below */

Thursday, August 30, 2018

Open Source Reinforcement Training

More open source resources here useful for training.    The frameworks involved here are likely the most instructive part.  And for tests to determine comparisons of speed and accuracy.  Note also the criticism of reinforcement learning, while conceptually a very powerful idea,  can be hard to calibrate for general problems.

Google releases open source reinforcement learning framework for training AI models
 Kyle Wiggers   @KYLE_L_WIGGERS  in Venture Beat

Reinforcement learning — an artificial intelligence (AI) technique that uses rewards (or punishments) to drive agents in the direction of specific goals — trained the systems that defeated Alpha Go world champions and mastered Valve’s Dota 2. And it’s a core part of Google subsidiary DeepMind’s deep Q-network (DQN), which can distribute learning across multiple workers in the pursuit of, for example, achieving “superhuman” performance in Atari 2600 games. The trouble is, reinforcement learning frameworks take time to master a goal, tend to be inflexible, and aren’t always stable.

That’s why Google is proposing an alternative: an open source reinforcement framework based on TensorFlow, its machine learning library. It’s available from Github starting today.  .... " 

No comments: