无知是福：通过信息门控实现鲁棒控制
Ignorance is Bliss: Robust Control via Information Gating
AI总结：
本文提出 InfoGating，通过可微信噪比门控学习任务所需最少信息，并在动力学、Q-learning 和行为克隆中提升控制策略的鲁棒泛化。
摘要：
信息简约性为学习表征提供了一种有用的归纳偏置，这类表征通过对噪声和虚假相关性保持鲁棒而实现更好的泛化。我们提出 information gating（信息门控），作为学习简约表征的一种方式，用以识别完成任务所需的最少信息。在门控信息时，我们可以学习揭示尽可能少的信息，使任务仍可求解；或者隐藏尽可能少的信息，使任务变得不可求解。我们使用信噪比的可微参数化来门控信息，该方法可应用于网络中的任意数值，例如在输入层擦除像素，或在某个中间层擦除激活。在输入层进行门控时，我们的模型会学习哪些视觉线索对给定任务是重要的。在中间层进行门控时，我们的模型会学习后续计算阶段需要哪些激活。我们将该方法称为 InfoGating。我们把 InfoGating 应用于多种目标，如多步前向和逆动力学模型、Q-learning 以及行为克隆，并强调 InfoGating 如何自然地帮助丢弃与控制无关的信息。结果表明，学习识别并使用最少信息可以改善下游任务中的泛化。基于 InfoGating 的策略对无关视觉特征明显更加鲁棒，从而改进 RL 模型的预训练和微调。
英文摘要：
Informational parsimony provides a useful inductive bias for learning representations that achieve better generalization by being robust to noise and spurious correlations. We propose \textit{information gating} as a way to learn parsimonious representations that identify the minimal information required for a task. When gating information, we can learn to reveal as little information as possible so that a task remains solvable, or hide as little information as possible so that a task becomes unsolvable. We gate information using a differentiable parameterization of the signal-to-noise ratio, which can be applied to arbitrary values in a network, e.g., erasing pixels at the input layer or activations in some intermediate layer. When gating at the input layer, our models learn which visual cues matter for a given task. When gating intermediate layers, our models learn which activations are needed for subsequent stages of computation. We call our approach \textit{InfoGating}. We apply InfoGating to various objectives such as multi-step forward and inverse dynamics models, Q-learning, and behavior cloning, highlighting how InfoGating can naturally help in discarding information not relevant for control. Results show that learning to identify and use minimal information can improve generalization in downstream tasks. Policies based on InfoGating are considerably more robust to irrelevant visual features, leading to improved pretraining and finetuning of RL models.
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。
