arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2004.00184cs.LGcs.CVstat.ML
收藏

生成模型外推的独立机制理论

A theory of independent mechanisms for extrapolation in generative models

Max Planck Institute for Intelligent Systems(马克斯·普朗克智能系统研究所);Max Planck Institute for Biological Cybernetics(马克斯·普朗克生物控制论研究所);ENS Rennes(雷恩高等师范学校)

Michel Besserve, Rémy Sun, Dominik Janzing, Bernhard Schölkopf

摘要:

生成模型可以被训练来模拟复杂的经验数据,但它们对于在以前未观察到的环境中进行预测有用吗?促进这种外推能力的直观想法是使这种模型的架构反映真实数据生成过程的因果图,使得可以独立于其他节点对每个节点进行干预。然而,该图的节点通常是不可观察的,导致过度参数化和缺乏因果结构的可识别性。我们开发了一个理论框架,以解决这一具有挑战性的情况下,通过定义一个较弱的形式的可识别性,基于机制的独立性原则。我们在玩具示例上证明,经典的随机梯度下降会阻碍模型的外推能力,这表明在训练过程中应明确执行机制的独立性。在真实世界数据上训练的深度生成模型上的实验支持了这些见解,并说明了如何利用这些模型的外推功能。

英文摘要:

Generative models can be trained to emulate complex empirical data, but are they useful to make predictions in the context of previously unobserved environments? An intuitive idea to promote such extrapolation capabilities is to have the architecture of such model reflect a causal graph of the true data generating process, such that one can intervene on each node independently of the others. However, the nodes of this graph are usually unobserved, leading to overparameterization and lack of identifiability of the causal structure. We develop a theoretical framework to address this challenging situation by defining a weaker form of identifiability, based on the principle of independence of mechanisms. We demonstrate on toy examples that classical stochastic gradient descent can hinder the model's extrapolation capabilities, suggesting independence of mechanisms should be enforced explicitly during training. Experiments on deep generative models trained on real world data support these insights and illustrate how the extrapolation capabilities of such models can be leveraged.

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。

↑