Salvato in:
| Autori principali: | , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2026
|
| Soggetti: | |
| Accesso online: | https://arxiv.org/abs/2605.17431 |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866910229177827328 |
|---|---|
| author | Hwang, Himchan Jeong, Hyeokju Chung, Gene Kim, Seungyeon Yoon, Sangwoong Park, Frank Chongwoo |
| author_facet | Hwang, Himchan Jeong, Hyeokju Chung, Gene Kim, Seungyeon Yoon, Sangwoong Park, Frank Chongwoo |
| contents | We propose MATE, a simple yet effective memory architecture for solving Contextual Markov Decision Processes (CMDPs), a family of MDPs parameterized by an unobserved context. In CMDPs, an optimal agent can adapt online by maintaining the posterior belief over contexts. MATE replaces this intractable posterior with a sum-aggregated memory, leveraging the posterior's permutation invariance to retain provably sufficient expressiveness. Compared to prior memory architectures, MATE avoids the growing per-step rollout cost of Transformers and the gradient issues commonly associated with Recurrent Neural Networks (RNNs). Extensive evaluations across diverse benchmarks demonstrate that MATE provides clear computational advantages while achieving performance comparable to standard sequence-model baselines. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_17431 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | MATE: Solving Contextual Markov Decision Processes with Memory of Accumulated Transition Embeddings Hwang, Himchan Jeong, Hyeokju Chung, Gene Kim, Seungyeon Yoon, Sangwoong Park, Frank Chongwoo Machine Learning Artificial Intelligence We propose MATE, a simple yet effective memory architecture for solving Contextual Markov Decision Processes (CMDPs), a family of MDPs parameterized by an unobserved context. In CMDPs, an optimal agent can adapt online by maintaining the posterior belief over contexts. MATE replaces this intractable posterior with a sum-aggregated memory, leveraging the posterior's permutation invariance to retain provably sufficient expressiveness. Compared to prior memory architectures, MATE avoids the growing per-step rollout cost of Transformers and the gradient issues commonly associated with Recurrent Neural Networks (RNNs). Extensive evaluations across diverse benchmarks demonstrate that MATE provides clear computational advantages while achieving performance comparable to standard sequence-model baselines. |
| title | MATE: Solving Contextual Markov Decision Processes with Memory of Accumulated Transition Embeddings |
| topic | Machine Learning Artificial Intelligence |
| url | https://arxiv.org/abs/2605.17431 |