Connectivity-contrastive learning: Combining causal discovery and representation learning for multimodal data
Connectivity-contrastive learning: Combining causal discovery and representation learning for multimodal data
复制标题
DOI：
--
复制
发表时间：
2023
期刊：
影响因子：
--
通讯作者：
H. Morioka;Aapo Hyvärinen
中科院分区：
文献类型：
--
作者：
H. Morioka;Aapo Hyvärinen
因果发现方法通常是基于每个节点的单变量观测来提取多个节点（变量）之间的因果关系。然而，人们经常遇到的情况是，每个节点是多元的，即有多个观测模式。此外，观测到的模态可能是通过未知的混合过程产生的，使得一些原始的潜在变量在节点内纠缠。在这种多式联运的情况下，现有的框架不能适用。为了分析这些数据，我们提出了一个新的因果表示学习框架，称为连接-对比学习（CCL）。CCL解开了观测混合，提取了一组相互独立的潜在成分，每个潜在成分在节点之间都有单独的因果结构。实际学习通过一种新的自监督学习方法进行，该方法的借口任务是根据节点对的观察结果预测一对节点的标签。我们提出的定理表明，在弱技术假设下，CCL确实可以识别潜在成分和多模态因果结构，直到一些不确定性。最后，我们通过实验证明了与最先进的基线相比，其优越的因果发现性能，特别是对潜在混杂因素的鲁棒性。
Causal discovery methods typically extract causal relations between multiple nodes (variables) based on univariate observations of each node. However, one frequently encounters situations where each node is multivariate, i.e. has multiple observational modalities. Furthermore, the observed modalities may be generated through an unknown mixing process, so that some original latent variables are entangled inside the nodes. In such a multimodal case, the existing frameworks cannot be applied. To analyze such data, we propose a new causal representation learning framework called connectivity-contrastive learning (CCL). CCL disentangles the observational mixing and extracts a set of mutually independent latent components, each having a separate causal structure between the nodes. The actual learning proceeds by a novel self-supervised learning method in which the pretext task is to predict the label of a pair of nodes from the observations of the node pairs. We present theorems which show that CCL can indeed identify both the latent components and the multimodal causal structure under weak technical assumptions, up to some in-determinacy. Finally, we experimentally show its superior causal discovery performance compared to state-of-the-art baselines, in particular demonstrating robustness against latent confounders.