EiHi Net: Out-of-Distribution Generalization Paradigm
It addresses a critical issue in AI for robust model deployment, but appears incremental as it builds on existing contrastive learning methods like SimCLR and VIC-Reg.
The paper tackles the out-of-distribution generalization problem in deep learning by developing the EiHi net, a new learning paradigm that improves performance on the challenging Nico dataset, achieving significant improvements over current state-of-the-art results without using domain information.
This paper develops a new EiHi net to solve the out-of-distribution (OoD) generalization problem in deep learning. EiHi net is a model learning paradigm that can be blessed on any visual backbone. This paradigm can change the previous learning method of the deep model, namely find out correlations between inductive sample features and corresponding categories, which suffers from pseudo correlations between indecisive features and labels. We fuse SimCLR and VIC-Reg via explicitly and dynamically establishing the original - positive - negative sample pair as a minimal learning element, the deep model iteratively establishes a relationship close to the causal one between features and labels, while suppressing pseudo correlations. To further validate the proposed model, and strengthen the established causal relationships, we develop a human-in-the-loop strategy, with few guidance samples, to prune the representation space directly. Finally, it is shown that the developed EiHi net makes significant improvements in the most difficult and typical OoD dataset Nico, compared with the current SOTA results, without any domain ($e.g.$ background, irrelevant features) information.