TE
TechEcho
Home24h TopNewestBestAskShowJobs
GitHubTwitter
Home

TechEcho

A tech news platform built with Next.js, providing global tech news and discussions.

GitHubTwitter

Home

HomeNewestBestAskShowJobs

Resources

HackerNews APIOriginal HackerNewsNext.js

© 2025 TechEcho. All rights reserved.

OpenAI Retro Contest

197 pointsby gdbabout 7 years ago

14 comments

mustdeparthastyabout 7 years ago
Two months to advance the state of the art on a complex physics-based game with branching paths. Current approaches such as DQNs or god forbid DRL[1] barely reach the performance of my three year old cousin in atari game score maximization and are mostly non-transferable to new levels... Good luck.<p>[1] <a href="https:&#x2F;&#x2F;www.alexirpan.com&#x2F;2018&#x2F;02&#x2F;14&#x2F;rl-hard.html" rel="nofollow">https:&#x2F;&#x2F;www.alexirpan.com&#x2F;2018&#x2F;02&#x2F;14&#x2F;rl-hard.html</a>
评论 #16769152 未加载
评论 #16770927 未加载
评论 #16768977 未加载
评论 #16771182 未加载
评论 #16771140 未加载
aquovaabout 7 years ago
I&#x27;m a big retro Sega fan, and I&#x27;ve always wanted to look into doing something like this, but this seems... really difficult. Would the best approach be to jump right in and hope for the best, or are there any sources I should look into?
评论 #16771560 未加载
评论 #16769677 未加载
minimaxirabout 7 years ago
The demo GIFs show Sonic 1 and 3, but all 3 Genesis games have slightly different physics and mechanics which could trip up an AI if not trained on all 3 games. Is the challenge just using the Sonic 1 engine?
评论 #16768776 未加载
评论 #16770779 未加载
chpmrcabout 7 years ago
I&#x27;m wondering if there&#x27;s a MOOC that takes you from zero to being able to build a system that learns how to play such a game, maybe focusing a bit less on the math (especially the proofs). I took Andrew Ng&#x27;s course on Coursera but I feel like the gap between what I know and what&#x27;s needed for this contest is huge. Am I wrong?
评论 #16770667 未加载
评论 #16783873 未加载
bagrowabout 7 years ago
Seems like a stretch to call this “transfer learning”. Maybe training on Sonic and testing on Mario.<p>Would be cool to see some kind of adversarial competition. You train to, say, beat a game level but you test to beat someone else’s submission. (Short on the specifics, I know.)
评论 #16778828 未加载
评论 #16773945 未加载
评论 #16771355 未加载
taericabout 7 years ago
Cynical prediction one: Nothing learned here will be readily transferrable to another domain. :(
rmellowabout 7 years ago
I expect training &amp; testing to make use of emulators &amp; ROMs - wouldn&#x27;t Sega potentially have a problem with this or is it considered fair use?
评论 #16769498 未加载
mlbossabout 7 years ago
Can we move away from video games ? I know games provide a closed loop for these kind of experiments&#x2F;contests, but I want to see more practical application of RL. How about code generation that create programs based on the test cases or RL agent that can use a 3d design software.
评论 #16768799 未加载
评论 #16769017 未加载
hanozabout 7 years ago
Couldn&#x27;t a new level contain some sort of new scenario that would be completely impossible to navigate based on previous experience without some <i>very</i> general AI?
mrfusionabout 7 years ago
What’s the input to the agent? Just all the pixels on the screen?
评论 #16768851 未加载
dnauticsabout 7 years ago
Is this a fair representation of a human playing a game? Usually as a human player I don&#x27;t go into a level expecting to clear it never having seen it before.
评论 #16769034 未加载
评论 #16778845 未加载
评论 #16771397 未加载
Mizzaabout 7 years ago
I couldn&#x27;t find - high score or fastest time? Totally different skills.
评论 #16770300 未加载
ikeboyabout 7 years ago
Do you win anything?
评论 #16768449 未加载
评论 #16768445 未加载
Ancalagonabout 7 years ago
Was this just released today?
评论 #16768872 未加载