New method explains AI decisions by accounting for how inputs depend on each other

Original: Computing Actual Causes for Neural Network Predictions under Structured Causal Inputs

Source: arxiv.org ↗