Hacker News new | past | comments | ask | show | jobs | submit login
Evolving Stable Strategies (otoro.net)
32 points by wei_jok on Dec 19, 2017 | hide | past | favorite | 1 comment



I wonder what happens when you simply backprop using experience replay in either a CNN or fully connected net. Just run a random neural net, and take "samples" (inputs + outputs) every 1s or so. After 30s get an error, optionally "discount" it over time, and run backprop.




Join us for AI Startup School this June 16-17 in San Francisco!

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: