World PulseNow
Newsletter

PrefPoE: Advantage-Guided Preference Fusion for Learning Where to Explore

arXiv — cs.LG•Wednesday, November 12, 2025 at 5:00:00 AM
PositiveArtificial Intelligence
The PrefPoE framework, introduced on November 12, 2025, represents a breakthrough in reinforcement learning, particularly in the area of exploration. Traditional methods often struggle with high variance and inefficient policy updates due to naive entropy maximization. PrefPoE addresses these issues by employing a Preference-Product-of-Experts approach, which intelligently guides exploration based on action advantages. This method not only stabilizes policy updates but also enhances training stability and sample efficiency. The framework has demonstrated impressive performance improvements across various control tasks, including a 321% increase on HalfCheetah-v4, a 69% increase on Ant-v4, and a 276% increase on LunarLander-v2. Unlike standard PPO, which can suffer from entropy collapse, PrefPoE maintains adaptive exploration dynamics, preventing premature convergence and ensuring more effective learning. This advancement highlights the importance of learning where to explore, which is …
— via World Pulse Now AI Editorial System
Read Original

Was this article worth reading? Share it

Recommended Readings
EvoLM: In Search of Lost Language Model Training Dynamics
arXiv — cs.LG4 hours ago
EvoLM: In Search of Lost Language Model Training Dynamics
PositiveArtificial Intelligence
EvoLM is a new model suite designed to analyze the training dynamics of language models (LMs) across various stages, including pre-training and fine-tuning. By training over 100 LMs with 1B and 4B parameters, EvoLM provides insights into the effectiveness of design choices and their impact on both language modeling and problem-solving capabilities. Key findings emphasize the diminishing returns of excessive pre-training and the importance of continued pre-training to mitigate forgetting during domain-specific tasks.
Read full article
via arXiv — cs.LG
Potent but Stealthy: Rethink Profile Pollution against Sequential Recommendation via Bi-level Constrained Reinforcement Paradigm
arXiv — cs.LG2 days ago
Potent but Stealthy: Rethink Profile Pollution against Sequential Recommendation via Bi-level Constrained Reinforcement Paradigm
PositiveArtificial Intelligence
The paper titled 'Potent but Stealthy: Rethink Profile Pollution against Sequential Recommendation via Bi-level Constrained Reinforcement Paradigm' addresses vulnerabilities in sequential recommenders, particularly to adversarial attacks. It highlights the Profile Pollution Attack (PPA), which subtly contaminates user interactions to induce mispredictions. The authors propose a new method called CREAT, which combines bi-level optimization with reinforcement learning to enhance the stealthiness and effectiveness of such attacks, overcoming limitations of previous methods.
Read full article
via arXiv — cs.LG
LDC: Learning to Generate Research Idea with Dynamic Control
arXiv — cs.CL2 days ago
LDC: Learning to Generate Research Idea with Dynamic Control
PositiveArtificial Intelligence
Recent advancements in large language models (LLMs) highlight their potential in automating scientific research ideation. Current methods often produce ideas that do not meet expert standards of novelty, feasibility, and effectiveness. To address these issues, a new framework is proposed that combines Supervised Fine-Tuning (SFT) and controllable Reinforcement Learning (RL) to enhance the quality of generated research ideas through a two-stage approach.
Read full article
via arXiv — cs.CL
Behaviour Policy Optimization: Provably Lower Variance Return Estimates for Off-Policy Reinforcement Learning
arXiv — cs.LG2 days ago
Behaviour Policy Optimization: Provably Lower Variance Return Estimates for Off-Policy Reinforcement Learning
PositiveArtificial Intelligence
The paper titled 'Behaviour Policy Optimization: Provably Lower Variance Return Estimates for Off-Policy Reinforcement Learning' addresses the challenges of high-variance return estimates in reinforcement learning algorithms. It highlights that well-designed behavior policies can collect off-policy data, leading to lower variance return estimates. This finding suggests that on-policy data collection is not optimal for variance, and the authors extend this insight to online reinforcement learning, where policy evaluation and improvement occur simultaneously.
Read full article
via arXiv — cs.LG
Thinker: Training LLMs in Hierarchical Thinking for Deep Search via Multi-Turn Interaction
arXiv — cs.CL2 days ago
Thinker: Training LLMs in Hierarchical Thinking for Deep Search via Multi-Turn Interaction
PositiveArtificial Intelligence
The article presents Thinker, a hierarchical thinking model designed to enhance the reasoning capabilities of large language models (LLMs) through multi-turn interactions. Unlike previous methods that relied on end-to-end reinforcement learning without supervision, Thinker allows for a more structured reasoning process by breaking down complex problems into manageable sub-problems. Each sub-problem is represented in both natural language and logical functions, improving the coherence and rigor of the reasoning process.
Read full article
via arXiv — cs.CL
From Efficiency to Adaptivity: A Deeper Look at Adaptive Reasoning in Large Language Models
arXiv — cs.CL2 days ago
From Efficiency to Adaptivity: A Deeper Look at Adaptive Reasoning in Large Language Models
NeutralArtificial Intelligence
Recent advancements in large language models (LLMs) have shifted the focus of reasoning as a benchmark for intelligence evaluation. This article critiques the uniform reasoning strategies employed by current LLMs, which often generate lengthy reasoning for simple tasks while struggling with complex ones. It introduces the concept of adaptive reasoning, which allows models to adjust their reasoning efforts based on task difficulty and uncertainty, and outlines three key contributions to understanding this approach.
Read full article
via arXiv — cs.CL
Harnessing Bounded-Support Evolution Strategies for Policy Refinement
arXiv — cs.LG2 days ago
Harnessing Bounded-Support Evolution Strategies for Policy Refinement
PositiveArtificial Intelligence
The article discusses the use of Triangular-Distribution Evolution Strategies (TD-ES) for refining robot policies through on-policy reinforcement learning (RL). It addresses challenges posed by noisy gradients and proposes a method that combines bounded triangular noise with a centered-rank finite-difference estimator. The two-stage process, involving PPO pretraining followed by TD-ES refinement, enhances success rates by 26.5% while reducing variance, making it a promising approach for improving robotic manipulation tasks.
Read full article
via arXiv — cs.LG
DiAReL: Reinforcement Learning with Disturbance Awareness for Robust Sim2Real Policy Transfer in Robot Control
arXiv — cs.LG2 days ago
DiAReL: Reinforcement Learning with Disturbance Awareness for Robust Sim2Real Policy Transfer in Robot Control
PositiveArtificial Intelligence
The paper titled 'DiAReL: Reinforcement Learning with Disturbance Awareness for Robust Sim2Real Policy Transfer in Robot Control' discusses the introduction of a disturbance-augmented Markov decision process (DAMDP) to enhance reinforcement learning in robotic control. It addresses the challenges of sim2real transfer, where models trained in simulation often fail to perform effectively in real-world scenarios due to discrepancies in system dynamics. The proposed approach aims to improve the robustness and stabilization of control responses in robotic systems.
Read full article
via arXiv — cs.LG