ai.hackcv
研究进展 · 学界与业界突破
全部生成式 AIAgent / 智能体多模态计算机视觉自然语言处理强化学习AI 安全编程助手公司与产品政策与监管研究进展AI 视频
论文精选 60已解读

Robust PAC Learning of Concurrent Stochastic Games· 并发随机游戏的鲁棒 PAC 学习

We introduce the first Probably Approximately Correct (PAC) learning framework for general…

AI 解读并发随机游戏的鲁棒 PAC 学习框架，解决纳什均衡存在性问题。
领域：cs.LG作者：Angel Y. He、David Parker
#cs.LG
📎 arXiv🕒 09-04 01:58🔗 arxiv.org
论文精选 60已解读

Parameterised graph theory for tensor networks: entanglement rerouting, structural simplification, and agnostic tomography· 参数化图论与张量网络

Parameterised graph theory studies how the complexity of graph-theoretic problems depends …

AI 解读参数化图论在张量网络中的应用，优化状态表示与学习复杂度
领域：quant-ph作者：Matthias C. Caro、Natalie McHugh、Sergii Strelchuk
#quant-ph
📎 arXiv🕒 09-04 01:50🔗 arxiv.org
论文精选 65已解读

RoboTTT: Context Scaling for Robot Policies· RoboTTT: 扩展机器人策略的上下文规模

Recent robot foundation models operate with single-step or short-history visuomotor contex…

AI 解读机器人模型 RoboTTT 实现了 8K 时间步的上下文扩展，大幅提升多阶段任务性能。
领域：cs.RO作者：Yunfan Jiang、Yevgen Chebotar、Ruijie Zheng
#cs.RO
📎 arXiv🕒 07-17 01:59🔗 arxiv.org
论文精选 65已解读

Decoding Market Emotion from Blockchain Activity: A Data-Driven Sentiment Classifier· 从区块链活动解码市场情绪：数据驱动的情绪分类器

The growing use of Bitcoin as a decentralized digital asset and investment tool has sparke…

AI 解读通过区块链交易、比特币历史价格和推特情绪分析来预测市场情绪的新方法。
领域：cs.LG作者：Arthur G. Bubolz、Abreu Quevedo、Giancarlo Lucca
#cs.LG
📎 arXiv🕒 07-17 01:52🔗 arxiv.org
论文精选 82

Data Driven Block Replacement Scheduling· 数据驱动的块替换调度

We develop data-driven algorithms for maintaining $N$ independent identical machines under…

领域：cs.LG作者：Aniruddhan Ganesaraman、VIdyadhar Kulkarni
#cs.LG
📎 arXiv🕒 07-17 01:31🔗 arxiv.org
论文精选 65已解读

Leveraging unlabelled data for generalizable neural population decoding· 利用未标记数据进行泛化神经群体解码

Robust and accurate neural decoders are integral to neurotechnologies such as brain-comput…

AI 解读利用未标记数据增强神经解码模型，提高泛化性能。
领域：cs.LG作者：Ximeng Mao、Nanda H. Krishna、Avery Hee-Woon Ryoo
#cs.LG
📎 arXiv🕒 07-16 01:58🔗 arxiv.org
论文精选 65已解读

Linear Independent Component Analysis via Optimal Transport· 通过最优传输实现线性独立成分分析

Linear Independent Component Analysis (ICA) recovers jointly independent source signals fr…

AI 解读通过最优传输距离优化线性独立成分分析，提高信号恢复准确性。
领域：cs.LG作者：Ashutosh Jha、Michel Besserve、Simon Buchholz
#cs.LG
📎 arXiv🕒 07-16 01:56🔗 arxiv.org
论文精选 65已解读

MetaPerch: Learning from metadata for bioacoustics foundation models· MetaPerch: 利用元数据的生物声学基础模型

Bioacoustic foundation models rely on large-scale citizen science platforms like Xeno-Cant…

AI 解读生物声学模型利用元数据提升物种识别性能。
领域：cs.LG作者：Mustafa Chasmai、Vincent Dumoulin、Jenny Hamer
#cs.LG
📎 arXiv🕒 07-16 01:42🔗 arxiv.org
论文精选 65已解读

Screening of Biosecurity Features in Metagenomic Data with Evo 2 Probes· 宏基因组数据中使用 Evo 2 探针筛选生物安全特征

Genomic foundation models such as Evo 2 learn rich sequence representations, but their val…

AI 解读使用 Evo 2 探针在宏基因组数据中筛查生物安全特征，如抗微生物耐药性。
领域：q-bio.GN作者：Jeremy Guntoro、Alexander Dack、Dylan Danno
#q-bio.GN
📎 arXiv🕒 07-16 01:38🔗 arxiv.org
论文精选 65已解读

The Seriality Gap in Video Diffusion Models· 视频扩散模型的串行性差距

When one ball strikes another, then another, video models should predict the consequences …

AI 解读视频扩散模型在因果链长时预测表现下降，串行计算结构需改进。
领域：cs.LG作者：Jorge Diaz Chao、Konpat Preechakul、Yuxi Liu
#cs.LG
📎 arXiv🕒 07-15 01:59🔗 arxiv.org
论文精选 65已解读

Ensemble Controlled-Flow Filtering for Implicit Data Assimilation· 隐式数据同化的集成控制流过滤方法

Data assimilation estimates the state of a dynamical system from model forecasts and incom…

AI 解读提出一种新的数据同化方法，适用于非高斯、多对一的隐式观测模型。
领域：stat.ML作者：Zhuoyuan Li、Yue Zhao、Ming Li
#stat.ML
📎 arXiv🕒 07-15 01:16🔗 arxiv.org
论文精选 65已解读

Robustness of Deep Learning Models for PV Power Forecasting under NWP Forecast Errors: A Spatiotemporal and Physically Interpretable Analysis· 光伏功率预测中深度学习模型的鲁棒性分析

Engineering use of AI forecasting models requires not only high nominal accuracy but also …

AI 解读分析深度学习模型在光伏功率预测中的鲁棒性
领域：physics.ao-ph作者：Dandan Chen、Yan Zhao、Xuepeng Chen
#physics.ao-ph
📎 arXiv🕒 07-15 00:48🔗 arxiv.org
论文精选 61已解读

Requential Coding: Pushing the Limits of Model Compression with Self-Generated Training Data· 重新序编码：利用自生成训练数据推动模型压缩极限

Compression is fundamental to intelligence. A model that can represent its training data a…

AI 解读通过自生成训练数据实现高效模型压缩，揭示了大规模模型对小规模模型的压缩优势。
领域：cs.LG作者：Shikai Qiu、Marc Finzi、Yujia Zheng
#cs.LG
📎 arXiv🕒 07-14 01:58🔗 arxiv.org
论文精选 61已解读

Input-Aware Dynamic Backdoor Attack Against Quantum Neural Networks· 针对量子神经网络的输入感知动态后门攻击

Quantum Neural Networks (QNNs) are a promising framework for quantum machine learning on n…

AI 解读针对量子神经网络提出首例输入感知动态后门攻击模型Q-DIBA。
领域：quant-ph作者：Junrui Zhang、Zemin Chen、Lusi Li
#quant-ph
📎 arXiv🕒 07-14 01:34🔗 arxiv.org
论文精选 61已解读

Relaxing Faithfulness with Intervention-Only Causal Discovery· 仅用干预的因果发现

Causal discovery algorithms learn a network that describes the causal dependencies among r…

AI 解读利用干预信息进行因果发现，放松对忠实性的要求。
领域：cs.LG作者：Bijan Mazaheri、Jiaqi Zhang、Caroline Uhler
#cs.LG
📎 arXiv🕒 07-14 01:12🔗 arxiv.org
论文精选 61已解读

An Exact Instrument for State Usage in Selective State-Space Models, and the Input-Driven Migration It Reveals· 状态空间模型中的精确测量工具及其输入驱动迁移

Selective state-space models such as Mamba route information through a bank of first-order…

AI 解读提出状态空间模型中模式使用的精确测量工具，揭示输入驱动的模式迁移现象。
领域：cs.LG作者：Raktim Bhattacharya
#cs.LG
📎 arXiv🕒 07-14 00:48🔗 arxiv.org
论文精选 60已解读

PHINN-EEG: Topological Time-Series Analysis of Dream-State EEG -- Dynamic Betti Curves for Dream Content Classification and Topology-Conditioned Neural Signal Synthesis

Current electroencephalography (EEG)-based dream detection relies on power spectral densit…

AI 解读引入 PHINN-EEG 模型，通过拓扑时间序列分析提升梦境内容分类性能。
领域：q-bio.NC作者：Ren Takahashi、Emre Yusuf、Jayabrata Bhaduri
#q-bio.NC
📎 arXiv🕒 07-11 01:59🔗 arxiv.org
论文精选 60已解读

Deep Gaussian Processes on Directed Acyclic Graphs

Many real-world processes can be represented as compositions of functions along a directed…

AI 解读提出在有向无环图上的深度高斯过程模型，以处理部分观测函数的挑战。
领域：stat.ML作者：Federico L. Perlino、Oliver Hamelijnck、Adam M. Johansen
#stat.ML
📎 arXiv🕒 07-11 01:41🔗 arxiv.org