arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1703.07718cs.LG
收藏

独立控制功能

Independently Controllable Features

McGill University(麦吉尔大学);École Polytechnique Fédérale de Lausanne(洛桑联邦理工学院);Université de Montréal(蒙特利尔大学)

Emmanuel Bengio, Valentin Thomas, Joelle Pineau, Doina Precup, Yoshua Bengio

摘要:

寻找能够解开真实数据中不同变化原因的特征是一项艰巨的任务,但在自然图像等静态领域却受到了相当大的关注。在交互式环境中,智能体可以有意识地采取行动,这为更好地处理这项任务提供了机会,因为智能体可以尝试不同的行动并观察其效果。我们引入的想法,在交互式环境中,潜在的因素,控制观察到的数据的变化,可以通过找出代理可以控制。我们提出了一个天真的方法来找到解释或衡量学习者的行动的效果的因素,并在说明性实验中进行测试。

英文摘要:

Finding features that disentangle the different causes of variation in real data is a difficult task, that has nonetheless received considerable attention in static domains like natural images. Interactive environments, in which an agent can deliberately take actions, offer an opportunity to tackle this task better, because the agent can experiment with different actions and observe their effects. We introduce the idea that in interactive environments, latent factors that control the variation in observed data can be identified by figuring out what the agent can control. We propose a naive method to find factors that explain or measure the effect of the actions of a learner, and test it in illustrative experiments.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑