Towards plausibility in time series counterfactual explanations

Marcin Kostrzewa, Krzysztof Galus, Maciej Zięba

We present a new method for generating plausible counterfactual explanations for time series classification problems. The approach performs gradient-based optimization directly in…