CIGA

CIGA is a framework designed to learn causally invariant representations from graph data, specifically addressing challenges in out-of-distribution generalization. It has applications in drug discovery, validating its relevance to molecular tools.