TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

Reinforcement Learning with Unsupervised Auxiliary Tasks

47 点作者 tonybeltramelli超过 8 年前

3 条评论

TFortunato超过 8 年前
Interesting. I&#x27;m not a deep learning guy, but from what I can gather, the new auxiliary tasks are to be rewarded for &quot;pixel changes&quot; and &quot;network features&quot;.<p>I haven&#x27;t nearly finished reading the paper, but is it safe to say this is similar (at a very high level), to a type of &quot;novelty search&quot;, where the agent is searching not only for a policy that is directly accomplishing the task at hand, but also for novel stimulus (in the case of pixel changes), and novel internal states (features, or maximally activated hidden nodes in the language of the paper), and that the benefit of this would be to more easily find relevant features that could be useful in the &quot;big picture&quot; task, and maybe not get as stuck in a non-optimal policy?<p>(I may be understanding this completely wrong...just an embedded guy looking to get more into this world, haha)
评论 #12979471 未加载
gallerdude超过 8 年前
I remember seeing a lecture on the importance of novelty in these kinds of things - good to see it applied.
deepnotderp超过 8 年前
Is this an attempt at unsupervised action decomposition or artificial curiosity?