Tag
The paper proposes Length-Aware Contrastive Learning for GUI Agents (LACL-GUI), a contrastive reinforcement learning framework that incorporates trajectory-level quality signals to improve agent performance by addressing reward-gradient misalignment.
TRACE introduces a novel approach to active 3D reconstruction by optimizing full sensor trajectories for ergodic coverage of scene information, outperforming next-best-view baselines with a 1.5 dB PSNR improvement.
The paper proposes locally straightening latent trajectories by maximizing cosine similarity between adjacent velocity vectors to reduce optimization difficulty in world model planning, with experiments on PushT dataset achieving 36% success rate for long-horizon planning.
ExTra introduces exploratory trajectory optimization for language model reinforcement learning, combining novelty rewards and entropy-guided prefix regeneration to improve both single-sample accuracy and inference-time coverage on mathematical reasoning benchmarks.
This paper proposes a CKM-driven framework for multi-UAV trajectory planning in urban inspection, using diffusion models to reconstruct high-fidelity channel quality maps and a graph attention network with soft actor-critic algorithm for communication-aware path planning.
This paper introduces CAPR (Cached-Amortized Path Refinement), a reinforcement learning algorithm for diffusion large language models that extracts tree-like supervision signals from the denoising trace without the compute cost of full tree rollouts. CAPR achieves state-of-the-art performance on reasoning benchmarks like GSM8K, Math500, Sudoku, and Countdown at roughly 0.75x the cost of flat rollouts.
This paper introduces FATE, an on-policy framework that leverages failure trajectories to enhance the safety and performance of tool-using LLM agents through self-evolution and Pareto-aware optimization.
OpenAI proposes POLO (Plan Online, Learn Offline), a framework combining model-based control with value function learning and coordinated exploration to enable efficient learning on complex control tasks like humanoid locomotion and dexterous manipulation with minimal real-world experience.
OpenAI introduces a method for learning complex nonlinear system dynamics using deep generative models over temporal segments, enabling stable long-horizon predictions and differentiable trajectory optimization for model-based control.