paper-with-me

Papers

Rethinking Transformers in Solving POMDPs

2024-05-27 · Chenhao Lu, Ruizhe Shi, Yuyao Liu, Kaizhe Hu, Simon S. Du, Huazhe Xu

Sequential decision-making algorithms such as reinforcement learning (RL) in real-world scenarios inevitably face environments with partial observability. This paper scrutinizes the effectiveness of a popular architecture, namely Transformers, in Partially Observable Markov Decision Processes (POMDPs) and reveals its theoretical limitations. We establish that regular languages, which Transformers struggle to model, are reducible to POMDPs. This poses a significant challenge for Transformers in learning POMDP-specific inductive biases, due to their lack of inherent recurrence found in other models like RNNs. This paper casts doubt on the prevalent belief in Transformers as sequence models for RL and proposes to introduce a point-wise recurrent structure. The Deep Linear Recurrent Unit (LRU) emerges as a well-suited alternative for Partially Observable RL, with empirical results highlighting the sub-optimal performance of the Transformer and considerable strength of LRU.

📄 PDF Abstract BibTeX arXiv:2405.17358

Code (1)

ctp314/tfporl 공식 구현 jax

Tasks

Decision MakingReinforcement Learning (RL)Sequential Decision Making

Methods 이 논문이 사용한 방법론

Attention 설명 없음
Linear Layer A Linear Layer is a projection $\mathbf{XW + b}$.
BPE Byte Pair Encoding, or BPE, is a subword segmentation algorithm that encodes rare and unknown words as sequences of subword units. The intuition is that various word…
Label Smoothing Label Smoothing is a regularization technique that introduces noise for the labels. This accounts for the fact that datasets may have mistakes in them, so maximizing the…
Adam 설명 없음
Residual Connection 설명 없음
Position-Wise Feed-Forward Layer 설명 없음
Multi-Head Attention 설명 없음

Similar Papers 제목 키워드 기반

Finite-State Controllers for (Hidden-Model) POMDPs using Deep Reinforcement Learning

2026-02-09 · David Hudák, Maris F. L. Galesloot, Martin Tappler, Martin Kurečka 외 arxiv

Solving partially observable Markov decision processes (POMDPs) requires computing policies under imperfect state information. Despite recent advances, the scalability of existing POMDP solvers remains limited. Moreover,…

Reinforcement Learning

Qualitative Possibilistic Mixed-Observable MDPs

2013-09-26 · Nicolas Drougard, Florent Teichteil-Konigsbuch, Jean-Loup Farges, Didier Dubois

Possibilistic and qualitative POMDPs (pi-POMDPs) are counterparts of POMDPs used to model situations where the agent's initial belief or observation probabilities are imprecise due to lack of past experiences or insuffic…

rho-POMDPs have Lipschitz-Continuous epsilon-Optimal Value Functions

2018-12-01 · NeurIPS 2018 12 · Mathieu Fehr, Olivier Buffet, Vincent Thomas, Jilles Dibangoye

Many state-of-the-art algorithms for solving Partially Observable Markov Decision Processes (POMDPs) rely on turning the problem into a “fully observable” problem—a belief MDP—and exploiting the piece-wise linearity and …

Monte Carlo Information-Oriented Planning

2021-03-21 · Vincent Thomas, Gérémy Hutin, Olivier Buffet

In this article, we discuss how to solve information-gathering problems expressed as rho-POMDPs, an extension of Partially Observable Markov Decision Processes (POMDPs) whose reward rho depends on the belief state. Point…

Solving infinite-horizon Dec-POMDPs using Finite State Controllers within JESP

2021-09-17 · Yang You, Vincent Thomas, Francis Colas, Olivier Buffet

This paper looks at solving collaborative planning problems formalized as Decentralized POMDPs (Dec-POMDPs) by searching for Nash equilibria, i.e., situations where each agent's policy is a best response to the other age…