cs.LG（2024-06-07）

📊 共 29 篇论文 | 🔗 4 篇有代码

🎯 兴趣领域导航

支柱二：RL算法与架构 (RL & Architecture) (13 🔗3) 支柱九：具身大模型 (Embodied Foundation Models) (12 🔗1) 支柱八：物理动画 (Physics-based Animation) (2) 支柱一：机器人控制 (Robot Control) (1) 支柱五：交互与反应 (Interaction & Reaction) (1)

🔬 支柱二：RL算法与架构 (RL & Architecture) (13 篇)

#题目一句话要点标签🔗⭐
1 Optimization of geological carbon storage operations with multimodal latent dynamic model and deep reinforcement learning 提出多模态潜在动态模型与深度强化学习方法，优化地质碳封存运营 reinforcement learning deep reinforcement learning SAC
2 MedualTime: A Dual-Adapter Language Model for Medical Time Series-Text Multimodal Learning MedualTime：一种用于医学时间序列-文本多模态学习的双适配器语言模型 contrastive learning multimodal ✅
3 Optimizing Automatic Differentiation with Deep Reinforcement Learning 提出基于深度强化学习的自动微分优化方法，显著减少雅可比矩阵计算中的乘法次数。 reinforcement learning deep reinforcement learning
4 Skill-aware Mutual Information Optimisation for Generalisation in Reinforcement Learning 提出技能感知互信息优化以解决强化学习泛化问题 reinforcement learning contrastive learning
5 Pretraining Decision Transformers with Reward Prediction for In-Context Multi-task Structured Bandit Learning 提出基于奖励预测的决策Transformer预训练方法，用于上下文多任务结构化Bandit学习。 decision transformer privileged information
6 Federated Representation Learning in the Under-Parameterized Regime 提出FLUTE算法，解决联邦表征学习在欠参数化场景下的性能瓶颈 representation learning distillation
7 Confidence-aware Contrastive Learning for Selective Classification 提出置信度感知的对比学习方法CCL-SC，提升选择性分类性能 contrastive learning
8 Stabilizing Extreme Q-learning by Maclaurin Expansion 提出基于麦克劳林展开的Extreme Q-learning，提升离线/在线强化学习稳定性。 reinforcement learning offline RL offline reinforcement learning
9 Reinforcement Learning and Regret Bounds for Admission Control 提出基于UCRL2的算法以优化M/M/c/S排队系统的接纳控制 reinforcement learning
10 On Minimizing Adversarial Counterfactual Error in Adversarial RL 提出对抗反事实误差(ACoE)以提升对抗强化学习的鲁棒性 reinforcement learning deep reinforcement learning DRL ✅
11 FlowMM: Generating Materials with Riemannian Flow Matching FlowMM：利用黎曼流匹配生成晶体材料，效率和灵活性均达到新高度 flow matching
12 Denoising-Aware Contrastive Learning for Noisy Time Series 提出Denoising-Aware对比学习(DECL)以提升含噪时间序列自监督学习性能 contrastive learning
13 Enhancing Size Generalization in Graph Neural Networks through Disentangled Representation Learning DISGEN：通过解耦表示学习增强图神经网络的尺寸泛化能力 representation learning ✅

🔬 支柱九：具身大模型 (Embodied Foundation Models) (12 篇)

#题目一句话要点标签🔗⭐
14 CorDA: Context-Oriented Decomposition Adaptation of Large Language Models for Task-Aware Parameter-Efficient Fine-tuning CorDA：面向上下文分解自适应的大语言模型参数高效微调方法 large language model instruction following
15 Bayesian Structural Model Updating with Multimodal Variational Autoencoder 提出基于多模态VAE的贝叶斯结构模型更新框架，提升小样本观测下的似然近似精度。 multimodal
16 CarbonSense: A Multimodal Dataset and Baseline for Carbon Flux Modelling CarbonSense：用于碳通量建模的多模态数据集与基线模型 multimodal
17 MMPolymer: A Multimodal Multitask Pretraining Framework for Polymer Property Prediction MMPolymer：多模态多任务预训练框架，提升聚合物性质预测精度。 multimodal
18 LinkGPT: Teaching Large Language Models To Predict Missing Links 提出LinkGPT，利用大型语言模型预测图数据中缺失的链接，实现高效图推理。 large language model
19 CTSyn: A Foundation Model for Cross Tabular Data Generation 提出CTSyn，一种基于扩散模型的跨表格数据生成基础模型，显著提升合成数据的质量和多样性。 foundation model
20 LogiCode: an LLM-Driven Framework for Logical Anomaly Detection LogiCode：一种基于LLM的逻辑异常检测框架，用于工业场景。 large language model
21 Spectrum: Targeted Training on Signal to Noise Ratio Spectrum：基于信噪比选择性训练加速大语言模型微调 large language model
22 The Factorization Curse: Which Tokens You Predict Underlie the Reversal Curse and More 揭示语言模型“分解诅咒”：探究Token预测目标对信息检索能力的影响 large language model
23 Hints-In-Browser: Benchmarking Language Models for Programming Feedback Generation 提出Hints-In-Browser，在浏览器端基准测试编程反馈生成语言模型，兼顾质量、成本、隐私和时间。 large language model
24 FunBO: Discovering Acquisition Functions for Bayesian Optimization with FunSearch FunBO：利用FunSearch发现贝叶斯优化中的新型采集函数 large language model
25 Helpful or Harmful Data? Fine-tuning-free Shapley Attribution for Explaining Language Model Predictions 提出FreeShap：一种免微调的Shapley值近似方法，用于解释语言模型预测 large language model ✅

🔬 支柱八：物理动画 (Physics-based Animation) (2 篇)

#题目一句话要点标签🔗⭐
26 Scaling up Probabilistic PDE Simulators with Structured Volumetric Information 提出基于有限体积法的概率偏微分方程模拟框架，提升不确定性建模的可扩展性。 spatiotemporal
27 Neural Laplace for learning Stochastic Differential Equations Neural Laplace框架扩展至随机微分方程学习，提升时空动态建模能力 spatiotemporal

🔬 支柱一：机器人控制 (Robot Control) (1 篇)

#题目一句话要点标签🔗⭐
28 Online Frequency Scheduling by Learning Parallel Actions 提出基于并行动作学习的在线频率调度方法，解决多用户MIMO系统资源分配问题。 sim-to-real reinforcement learning

🔬 支柱五：交互与反应 (Interaction & Reaction) (1 篇)

#题目一句话要点标签🔗⭐
29 VTrans: Accelerating Transformer Compression with Variational Information Bottleneck based Pruning VTrans：基于变分信息瓶颈的Transformer剪枝加速压缩 ReMoS

⬅️ 返回 cs.LG 首页 · 🏠 返回主页