Beating Atari with Natural Language Guided Reinforcement Learning — mentioned by yorwba

Mon, 02 Jul 2018 12:18:18 GMT

> NNs learn classification on a particular representation only To perform the task based on the description, you need to have learned to associate descriptions of images with the image itself. To train that ability you'll need lots of paired examples in both representations. That the description is not in the same representation as the image you want to recognize doesn't change the fact that both need to use a representation that has been trained on. Cross-representation learning is still useful because it increases the range of usable data (e.g. it can improve game-playing agents by telling…

Paper and discussion history. A mention is not an endorsement.

Beating Atari with Natural Language Guided Reinforcement Learning — HN submission

Thu, 20 Apr 2017 01:23:41 GMT

Beating Atari with Natural Language Guided Reinforcement Learning [pdf]

82 points; 12 discussion replies on HN. Paper and source history. A submission is not an endorsement.

Beating Atari with Natural Language Guided Reinforcement Learning · HackerPapers discussions

Beating Atari with Natural Language Guided Reinforcement Learning — mentioned by yorwba

Beating Atari with Natural Language Guided Reinforcement Learning — HN submission