We propose a Target Conditioned Representation Independence (TCRI) objective for domain generalization. TCRI addresses the limitations of existing domain generalization methods due to incomplete constraints. Specifically, TCRI implements regularizers motivated by conditional independence constraints that are sufficient to strictly learn complete sets of invariant mechanisms, which we show are necessary and sufficient for domain generalization. Empirically, we show that TCRI is effective on both synthetic and real-world data. TCRI is competitive with baselines in average accuracy while outperforming them in worst-domain accuracy, indicating desired cross-domain stability.
翻译:我们提出了一个目标性有条件代表独立(TCRI)目标,以概括化为目的,TCRI处理现有领域一般化方法因不完全限制而受到限制的问题,具体来说,TCRI实施基于有条件独立限制的正规化者,这些限制足以严格地学习全套变数机制,我们表明这些机制对于领域一般化是必要和充分的。我们经常地表明,TRIC在合成数据和现实世界数据上都是有效的。TCRI具有平均准确性基准的竞争力,而其表现优于最差的准确性,表明预期的跨领域稳定性。