The paper identifies an informational curse of horizon in goal-conditioned policy learning, showing that longer goal relabeling horizons during training can hurt policy generalization and performance. Controlled experiments with oracle planners separate training horizons from test horizons, revealing that behavioral cloning policies degrade with longer horizons while reinforcement learning mitigates this effect. The authors attribute this to reduced conditional mutual information between actions and hindsight-relabeled goals, and demonstrate that distilling short-horizon policy Jacobians into long-horizon policies improves performance, especially in combinatorial manipulation tasks.
arXiv:2608.30406v1 Announce Type: new Abstract: Goal-conditioned reinforcement learning struggles with long horizons when rewards are sparse. While a planner can provide subgoals to guide a low-level...
arXiv:2602. 05459v2 Announce Type: replace Abstract: Offline goal-conditioned reinforcement learning (GCRL) is typically benchmarked by the best tuned success rate of each method.
arXiv:2609.37119v1 Announce Type: cross Abstract: Recent approaches to reinforcement learning (RL) post-training for large language models increasingly remove the critic to reduce training instabilit...
The paper introduces MI‑SARSA, an on‑policy temporal‑difference algorithm that incorporates mutual‑information regularization to model bounded rationality in reinforcement learning. By penalizing state‑specific deviations from a learned marginal action prior, the algorithm selectively uses state information only when the expected return outweighs the informational cost, yielding a reward‑complexity tradeoff. MI‑SARSA also predicts reaction times, showing that stronger information penalties lead to simpler policies, lower control costs, and faster responses, while regularization mitigates performance loss after environmental shifts at the expense of asymptotic return.
arXiv:2509. 03206v2 Announce Type: replace-cross Abstract: Learning from reward functions and imitation learning of demonstrations are the two principal approaches for training autonomous systems that interact with an environment through action and observation.
Limiting‑Kernel Q(λ) (LKQL) is an off‑policy value estimator that blends n‑step truncation with a long‑horizon approximation based on the limiting kernel. It maintains the computational efficiency of n‑step methods while improving policy evaluation accuracy, especially for long‑horizon tasks. The authors prove faster convergence of LKQL’s operator under aperiodicity and near‑on‑policy conditions, and demonstrate empirical gains on MuJoCo continuous‑control benchmarks.
The paper introduces Generalized Implicit Temporal Abstraction (GITA), a method for goal-conditioned reinforcement learning that conditions a single value function on multiple temporal abstraction levels (k). By aggregating advantage-weighted supervision across various k values, GITA preserves both long-range signal and local resolution without committing to a single k. Experiments on OGBench show that GITA outperforms existing offline GCRL baselines, improving average success rates by 25 percentage points over HIQL and 7 percentage points over OTA.
The paper introduces the Dual-Latent World Model (Dual-WM), which separates local execution and long-range planning into distinct latent spaces and dynamics models. A new learning method, Long-Horizon Representation Learning with Weighted Rollout (LoRe), supervises predictions at both levels using exponential horizon weights. Experiments on five goal-conditioned visual control tasks show that Dual-WM improves success rates over strong baselines, especially at longer horizons.
arXiv:2609.24380v1 Announce Type: new Abstract: RLVR has substantially improved the reasoning capabilities of LLMs. However, existing methods typically parameterize temporal progression in the Markov...
In value-based reinforcement learning, improving the accuracy of policy evaluation has been shown to improve downstream policy optimization performance. The widely adopted family of approximations rel...
The paper investigates the problem of sharing a single critic across multiple parallel environments in reinforcement learning. It shows that when environments assign different expected returns to the same state, a shared critic must reconcile conflicting value targets, which can distort advantage estimates and misguide policy updates. The authors propose a simple fix—providing the critic with the environment index—demonstrating through bandit models and experiments on CartPole, MuJoCo, BipedalWalker, and 16 Procgen games that this conditional critic stabilizes learning and boosts returns, achieving a 40.8% improvement in aggregate normalized return on unseen levels.
The paper presents a constrained command‑conditioned Proximal Policy Optimization (PPO) executor for the MicroRTS real‑time strategy environment, enabling a single execution policy to follow diverse strategic commands. A Thompson‑sampling bandit strategist selects command tuples based on in‑game observations of the opponent’s strategy, rather than relying on opponent identity. In controlled experiments, the strategist‑executor system outperforms a flat PPO baseline against three of the four strongest opponents on a training map, including two held‑out opponents, while showing no significant difference against the remaining opponents.