EiHi Net: Out-of-Distribution Generalization Paradigm

TL;DR

EiHi Net introduces a flexible paradigm for out-of-distribution generalization in deep learning by combining contrastive learning with causal relationship establishment and human-in-the-loop pruning, significantly improving performance on challenging OoD datasets.

Contribution

The paper presents EiHi Net, a novel OoD generalization framework that fuses SimCLR and VIC-Reg, and incorporates human-in-the-loop pruning to enhance causal feature learning.

Findings

Achieves significant improvements on the Nico OoD dataset.

Effectively suppresses pseudo correlations between features and labels.

Demonstrates robustness without relying on domain-specific information.

Abstract

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.