A fear-inspired intrinsic reward, computed by a Siamese memory-augmented network over state sequences, helps a PPO agent avoid terminal states in MiniGrid Sidewalk without directly sampling them, though success rates remain near zero.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.AI 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Avoiding Death through Fear Intrinsic Conditioning
A fear-inspired intrinsic reward, computed by a Siamese memory-augmented network over state sequences, helps a PPO agent avoid terminal states in MiniGrid Sidewalk without directly sampling them, though success rates remain near zero.