| 1 |
Aero-VPA: Few-Step Flow-Matching Control over Non-Reconstructive JEPA Latents for Agile Flight on Edge Hardware |
提出Aero-VPA以解决边缘硬件上的快速飞行控制问题 |
flow matching JEPA vision-language-action |
✅ |
|
| 2 |
Density-Aware Reward Scaling: A Reinforcement Learning Framework for Reward-Proportional Sampling |
提出密度感知奖励缩放以解决强化学习中的奖励多样性问题 |
reinforcement learning multimodal |
|
|
| 3 |
OmniST: Topology-Invariant Spatiotemporal Forecasting with Guaranteed-Coverage Uncertainty at Unseen Sensor Locations |
提出OmniST以解决时空预测中传感器拓扑依赖问题 |
MAE spatiotemporal |
|
|
| 4 |
Stabilizing On-Policy Targeted Guidance in Agentic Reinforcement Learning |
提出目标引导框架以解决强化学习中的奖励优化问题 |
reinforcement learning distillation |
|
|
| 5 |
Training Specialist LLMs on Your Phone: Forward-Only Optimization via Block-Spectral Coordinates |
提出MobileSpec以解决手机端LLM训练效率低下问题 |
reinforcement learning large language model |
|
|
| 6 |
Cooperative Multi-Agent Reinforcement Learning with Submodular Trajectory Rewards |
提出一种新算法以优化多智能体强化学习中的轨迹奖励问题 |
reinforcement learning |
|
|
| 7 |
CANDO: Conservative Action-Noise Directional Optimization for Residual Reinforcement Learning |
提出CANDO以优化残差强化学习中的动作噪声适应问题 |
reinforcement learning |
✅ |
|
| 8 |
Aligning Training and Rollout Policies for End-to-End FP4 RL on Mixture-of-Experts LLMs |
提出MoeResQ以解决MoE模型中FP4 RL训练与推理不一致问题 |
reinforcement learning large language model |
|
|
| 9 |
Found by Search, Broken by a Shortcut: Transfer and Tampering of Searched Intrinsic Rewards |
提出一种搜索驱动的内在奖励机制以提升强化学习的迁移能力 |
reinforcement learning PPO |
|
|
| 10 |
Koopman Generative Operators for Efficient Probabilistic Time-Series Forecasting |
提出Koopman生成算子以解决高效概率时间序列预测问题 |
flow matching latent dynamics |
|
|
| 11 |
Self-Privileged Fine-Tuning |
提出自特权微调以提升语言模型的任务适应性 |
reinforcement learning privileged information |
|
|