Neurons in DRL agents are matched to short Boolean formulas over hand-defined state predicates, with anecdotal perturbation evidence that these matches reflect real behavior.
Interpretability beyond feature attribution: Quantitative testing with concept activation vectors (tcav)
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Compositional Concept-Based Neuron-Level Interpretability for Deep Reinforcement Learning
Neurons in DRL agents are matched to short Boolean formulas over hand-defined state predicates, with anecdotal perturbation evidence that these matches reflect real behavior.