Regret-Aware Policy Optimization: Environment-Level Memory for Replay Suppression under Delayed Harm
Fuente:
arXiv
Guardado en:
| Autor principal: | |
|---|---|
| Formato: | Preprint |
| Publicado: |
2026
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866918435624058880 |
|---|---|
| author | Hiremath, Prakul Sunil |
| author_facet | Hiremath, Prakul Sunil |
| contents | Safety in reinforcement learning (RL) is typically enforced through objective shaping while keeping environment dynamics stationary with respect to observable state-action pairs. Under delayed harm, this can lead to replay: after a washout period, reintroducing the same stimulus under matched observable conditions reproduces a similar harmful cascade.
We introduce the Replay Suppression Diagnostic (RSD), a controlled exposure-decay-replay protocol that isolates this failure mode under frozen-policy evaluation. We show that, under stationary observable transition kernels, replay cannot be structurally suppressed without inducing a persistent shift in replay-time action distributions.
Motivated by platform-mediated systems, we propose Regret-Aware Policy Optimization (RAPO), which augments the environment with persistent harm-trace and scar fields and applies a bounded, mass-preserving transition reweighting to reduce reachability of historically harmful regions.
On graph diffusion tasks (50-1000 nodes), RAPO suppresses replay, reducing re-amplification gain (RAG) from 0.98 to 0.33 on 250-node graphs while retaining 82\% of task return. Disabling transition deformation only during replay restores re-amplification (RAG 0.91), isolating environment-level deformation as the causal mechanism. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2604_07428 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Regret-Aware Policy Optimization: Environment-Level Memory for Replay Suppression under Delayed Harm Hiremath, Prakul Sunil Machine Learning Artificial Intelligence I.2.6; I.2.8 Safety in reinforcement learning (RL) is typically enforced through objective shaping while keeping environment dynamics stationary with respect to observable state-action pairs. Under delayed harm, this can lead to replay: after a washout period, reintroducing the same stimulus under matched observable conditions reproduces a similar harmful cascade. We introduce the Replay Suppression Diagnostic (RSD), a controlled exposure-decay-replay protocol that isolates this failure mode under frozen-policy evaluation. We show that, under stationary observable transition kernels, replay cannot be structurally suppressed without inducing a persistent shift in replay-time action distributions. Motivated by platform-mediated systems, we propose Regret-Aware Policy Optimization (RAPO), which augments the environment with persistent harm-trace and scar fields and applies a bounded, mass-preserving transition reweighting to reduce reachability of historically harmful regions. On graph diffusion tasks (50-1000 nodes), RAPO suppresses replay, reducing re-amplification gain (RAG) from 0.98 to 0.33 on 250-node graphs while retaining 82\% of task return. Disabling transition deformation only during replay restores re-amplification (RAG 0.91), isolating environment-level deformation as the causal mechanism. |
| title | Regret-Aware Policy Optimization: Environment-Level Memory for Replay Suppression under Delayed Harm |
| topic | Machine Learning Artificial Intelligence I.2.6; I.2.8 |
| url | https://arxiv.org/abs/2604.07428 |