Keeping a city running means making decisions that affect residents every day — from when to repair a road to how to use energy more efficiently. Across the EMEA region, cities and local software, infrastructure and integration partners are using NVIDIA accelerated computing to build physical AI agents and digital twins that help teams make […]
Across industries, the U.K. is putting the full stack to work — and the scale of deployment is growing fast, from enterprises transforming their businesses to startups solving some of the planet’s hardest problems. NVIDIA is putting that principle into practice, including through a five-year, $250 million compute commitment to strengthen the U.K.’s agentic safety […]
了解 Google DeepMind 和 GovTech Singapore 如何利用轻量级端侧 Gemma 模型，在不向云端传输任何数据的情况下，拦截私人聊天中的对话式诈骗。
了解 AWS Generative AI Innovation Center Partner Agent Factory 和 Anchor Browser 如何构建能够自主支付费用以访问受 x402 保护的网络资源和 API 的 AI 智能体。Amazon Bedrock AgentCore payments 可检查预算、签署交易并重试请求，而无需向智能体运行时暴露钱包凭证。
Higgsfield 借助 GPT-6 Astra 为创作者带来更长的视频，借助 GPT-Image-2.5 实现精准视觉效果，并借助 GPT-Live 开辟语音电影制作之路。
在数十或数百个 AWS 账户之间管理 AI 模型访问权限时，如果每个账户都必须单独接受 AWS Marketplace Private Pricing Offers，就会带来摩擦。本文介绍了在 AWS License Manager 中激活 MPPO 授权的两种方式：一种是用于一次性修复的轻量级脚本，另一种是用于持续事件驱动处理的 CDK 自动化方案。
Grok 4.7 现已在 Amazon Bedrock 上可用。在本文中，我们基于 Grok 4.7 构建一个工具调用智能体，并在四个推理力度级别下对其在三项推理任务上的表现进行评估，同时衡量每个已完成任务的正确性和令牌成本。我们将展示如何针对每个请求调整这种权衡，并在生产环境中对其进行监控。
ShipStation Global's engineering team built Deckhand, an AI IT help desk agent for Slack on Amazon Bedrock AgentCore. It answers routine IT questions from a curated knowledge base, reads error-message screenshots, and files Jira tickets. In its first three months it resolved 97.5% of channel queries with a 96% positive feedback rating.
Orca Security 构建了 Discovery AI，这是一个基于 Amazon Bedrock 的自然语言查询代理，可将日常英语问题转换为 Sonar 查询。本文分享了其多阶段架构及结果：与最初的基线相比，查询成本降低了 90%，输入令牌减少了 93%，延迟降低了 75%。
我们让开发者、企业、科学家和消费者都能使用 Gemini 4 Argon，这是我们最新的前沿模型。
AWS 工程师分享了在 Amazon SageMaker AI 上针对融合式专家混合（MoE）模型的内核优化方案：通过 Swap-AB 提升 WGMMA 利用率，利用 Triton 后端路由实现 FP8 量化，并消除 Triton JIT 重新编译。综合应用这些优化后，在 H200 GPU 上吞吐量最高提升 10%，首个令牌时间最高缩短 9%。
NVIDIA RTX Spark 为 PC 开启全新篇章——通过革新，将游戏 PC、创意工作室和个人 AI 超级计算机的能力融于一颗超级芯片之中。凭借高达 128GB 的统一内存和每秒 1 千万亿次浮点运算的 AI 算力，RTX Spark 提供 NVIDIA RTX 的全栈能力以及 […]
法国企业正在航天、金融服务、公共部门和科学研究等各行业部署智能体计算，展现出该国生态系统如何从雄心迈向实际应用。法国的进展建立在精心布局的基础之上。2025 年 2 月举行的人工智能行动峰会，以及 2025 年和 2026 年举行的“选择法国”峰会，促成了重大的算力投入承诺 […]
Learn how to run OpenAI GPT-6 Sol, Luna, and Astra on Amazon Bedrock through global cross-Region inference from Singapore, Jakarta, Thailand, Malaysia, and Taipei. This post covers the Converse and OpenAI-compatible APIs, IAM and SCP permissions, quota sizing, and monitoring inference from your source Region.
StudentSim 是一个研究系统，用于在 AI 辅导工具与学习者进行测试之前对其进行评估和改进。它模拟个人可能与辅导工具互动的方式，为改进辅导工具在国际象棋、写作和数学等学科中的表现提供反馈。文章《模拟学生学习，打造更优的 AI 辅导》最初发表于 Microsoft Research。
了解如何在 Amazon EKS 上使用多节点 Ray LLM Deep Learning Containers，将一个大型语言模型部署到两个 GPU 节点上，并在一个兼容 OpenAI 的端点后配置独立的预填充节点和解码节点。预构建容器附带完整的推理技术栈，因此您只需配置部署，无需自行组装。
了解如何使用 llama.cpp 在 Amazon SageMaker AI 上部署决策模型。本技术教程将通过用于 llama.cpp 的 AWS Deep Learning Container 提供兼容 OpenAI 的聊天补全服务，然后使用 SemIf 扩展运行时，以便在 CPU 或 GPU 上返回针对已声明选项的概率。
随着 AI 工厂从兆瓦级发展到吉瓦级，其建设所需已不止于先进的计算技术。开放标准正帮助业界更快速、更高效、更可靠地扩展 AI 基础设施。在 San Jose McEnery Convention Center 举行的 Open Compute Project（OCP）全球峰会上，NVIDIA 宣布了新的合作伙伴关系及贡献，旨在 […]
arXiv:2609.38880v2 Announce Type: replace-cross Abstract: We study the last iterate of standard importance-weighted temporal-difference (TD) learning from a single trajectory of a finite Markov decision process. The trajectory follows a behavior policy $b$ with an irreducible, possibly periodic state chain; a target policy $\pi$ covered by $b$ is evaluated with action importance ratios bounded by $\varrho$. Write $H=(1-\gamma)^{-1}$, with $\mu_{\min}$ the minimum stationary probability and $\mathfrak h_\star$ the largest expected hitting time of the reverse behavior state chain. We prove that the last iterate achieves sup-norm error at most $\varepsilon$ with high probability after $\widetilde O(\varrho H^3/(\mu_{\min}\varepsilon^2)+\mathfrak h_\star)$ transitions for $0<\varepsilon\le1$. The statistical term retains the cubic effective-horizon dependence of synchronous TD, while the initialization cost is additive in $\mathfrak h_\star$. Setting $b=\pi$ and $\varrho=1$ recovers the on-policy rate. For aperiodic behavior chains, $\mathfrak h_\star=\widetilde O(t_{\rm mix}/\mu_{\min})$ recovers a mixing-time bound. The proof combines an anchored local Poisson equation in reverse time to control fluctuations without a mixing-time factor, a compensation argument for successive hitting costs, and an expected cumulative propagation mass bound for decreasing step sizes. Given the policy ratios and their bound, an empirical state-count rule attains the same rate without prior knowledge of model parameters. The framework covers nonnegative affine importance-weighted prediction updates, including successor features, transition-dependent discounting, and coupled finite-mode predictions. A perturbation estimate yields a time-uniform guarantee for online return-variance estimation. A three-state, two-action construction gives matching minimax lower bounds for both terms up to logarithmic factors on the considered parameter classes.
arXiv:2609.38345v2 Announce Type: replace-cross Abstract: Multi-agent coding systems are designed to tackle complex software engineering tasks through collaboration. However, existing evaluations typically assume configured organizations are followed faithfully, whereas reality differs. This behavioral gap, combined with differences in underlying system components, prevents clear attribution of observed gains. To this end, we introduce OpenCollab, a multi-agent coding framework that provides a unified infrastructure for programmable collaboration and controllable runtime. Specifically, OpenCollab unifies organization design, enforces experimental control on a shared runtime, and tracks execution through fine-grained event streams. On this basis, we define Adherence to quantify whether the declared organization is actually realized. Our experiments reveal that agents collaborate very differently across configurations: changing any single dimension shifts Adherence, from 47.2% to as high as 97.2%. Furthermore, extensive agentic coding benchmarks show that OpenCollab (Duo) establishes new SOTA performance compared to the mainstream harnesses such as Mini-SWE-agent, OpenHands, Codex CLI, and Claude Code, showing that a well-designed organization can outperform strong existing harnesses, while OpenCollab's single-agent configuration uses the fewest tokens across all evaluated suites. OpenCollab establishes a unified multi-agent infrastructure for easy programmable collaboration and controlled causal evaluation.
arXiv:2609.38263v2 Announce Type: replace-cross Abstract: Feature selection in neural networks remains a challenging problem, particularly in the presence of noisy or contaminated data. LassoNet is a recent approach that addresses this issue by combining neural networks with hierarchical sparsity constraints, enabling simultaneous prediction and variable selection. However, its standard formulation relies on the mean squared error (MSE) loss, which is known to be highly sensitive to outliers. In this paper, we present Robust LassoNet, an extension of LassoNet that incorporates robust loss functions, such as Huber, Cauchy, Tukey's bisquare, and Nonnegative Garrote, to mitigate the effect of extreme observations. The proposed approach preserves the original optimization framework while improving stability under data contamination. Through experiments on synthetic and real datasets, we show that robust LassoNet significantly improves both predictive performance and feature selection accuracy in the presence of outliers or heavy tailed noise, while maintaining comparable performance in clean settings. These results highlight the importance of robustness in neural network based feature selection and suggest practical guidelines for choosing appropriate loss functions.
arXiv:2609.33967v2 Announce Type: replace-cross Abstract: Practical assistive and rehabilitative brain--computer interfaces require subject-independent motor-imagery EEG (MI-EEG) decoders that generalize to new users under limited target-user data and constrained compute. However, held-out-subject performance can be overstated when test-subject information influences preprocessing, model selection, or ensemble selection. We present \textit{ThinkNet}, a validation-controlled framework that combines train-only normalization, validation-guided evolutionary search, and validation-gated inference to identify compact decoders and inference policies for held-out subjects. We evaluate four-class BCI Competition IV-2a (session T) decoding with nine Leave-One-Subject-Out (LOSO) folds, three seeds, seven fixed decoder entries, and a broader search over ten representative decoder families; the held-out subject is never used for normalization, hyperparameter, architecture, or ensemble-policy selection. In the fixed benchmark, the validation-selected compact decoder achieved 44.35$\pm$15.41\% accuracy with 4.9K parameters, 19 KB FP32 weights, and 0.99 ms batch-1 Orin CUDA inference. Across the broader search, compact models ($\leq$25K parameters) achieved higher mean held-out accuracy than mid-size and large alternatives after selected retraining (40.10\% vs. 35.09\% and 34.78\%). Validation-gated ensembling improved over validation-selected single-model inference, reaching 43.98$\pm$16.25\% in the fixed benchmark and 43.31$\pm$15.88\% for the compact six-family ensemble. A non-deployable oracle analysis revealed a 6.1-point family-selection gap and near-zero validation--test correlation, showing that validation reliability remains a key bottleneck under subject shift. Thus, ThinkNet is a validation-controlled framework for compact MI-EEG model and inference-policy selection, rather than a single-architecture benchmark.
arXiv:2609.19844v2 公告类型：replace-cross 摘要：AI 生成的 RTL 验证计划可能满足服务提供方的模式要求，却在进入可信执行环节的边界处失效。我们提出 SecTB-RTL，这是一个可审计的框架，涵盖 31 项任务和 124 个人工编写的硬件安全回归测试。在逐步提高资源限制的条件下，确定性的非 AI 基线分别杀死了 36、75 和 78 个变异体。首次确认性运行（C1-R2）在模型执行前便告失败，原因是服务提供方拒绝了其响应模式。在未查看结果的情况下仅修复模式后，一次单独冻结配置的后续运行（C1-R3）完成了 1,860 次调用。服务提供方接受了 1,857 个响应，但只有 9 个通过了生产环境语义验证器。生成规则与执行规则并不匹配。因此，我们将此次运行保留为一起测量工具验证事件，不报告任何提示词效应估计。该事件表明，服务提供方接受响应或响应符合模式，并不能证明其执行有效性。编译和覆盖率仅是诊断手段；实际保存的产物必须通过完整的生产执行流程。此后的一次跟进运行因未满足预注册的证据完整性门槛而被排除，仅作为未来工作处理。我们发布了该基准、保留失败信息的契约、事件溯源记录，以及防止将基础设施行为误报为模型行为所需的治理控制措施。
arXiv:2609.16464v2 公告类型：replace-cross 摘要：孤独症谱系障碍（ASD）的临床管理在早期筛查方面面临瓶颈，主要原因是受过培训的专科人员稀缺，且传统评估工具具有主观性。在此，我们提出 ASDchat，这是一种专为循证 ASD 筛查设计的多模态大语言模型，以视频、音频和对话作为输入。ASDchat 采用双分支架构，其中决策分支生成筛查概率，证据分支生成可追溯、带时间戳且与孤独症诊断观察量表（ADOS-2）相对应的行为证据。该模型在来自中国 27 个研究中心、包含 1,035 名参与者的数据集上进行了训练和评估，参与者涵盖典型发育（TD）儿童、ASD 儿童及患有其他发育障碍的儿童。在区分 ASD 与 TD 时，ASDchat 的受试者工作特征曲线下面积（AUC）达到 0.953 $\pm$ 0.021。在未用于训练的 6 个留出研究中心上，平均 AUC 为 0.961。此外，对行为维度进行无监督聚类，将 ASD 病例划分为六种具有不同表型特征的亚型，ASDchat 为每种亚型提出相应的干预建议。ASDchat 为临床实践中开展大规模循证 ASD 筛查提供了一条可行路径。
arXiv:2609.16082v2 Announce Type: replace-cross Abstract: Social media platforms are popular channels for disseminating information, owing to their large user bases and ease of access. Companies also use social media as an important aspect of the advertising process. By creating high-quality posts, companies can strengthen their engagement metrics and increase their follower count. While a growing body of research has examined social media engagement, fewer studies have jointly examined the visual, textual, and temporal features of image posts, even though these features collectively determine the performance of content on social media. To understand the important drivers of social media engagement, we collect image posts of furniture firms on Facebook and extract visual, temporal, and textual features from them using text and image analytics methods. We evaluate several machine learning models--including Random Forest, Light Gradient Boosting Machine (LightGBM), and eXtreme Gradient Boosting (XGBoost)--to assess the drivers and the prediction power of social media engagement using the features from our data. Our primary model, LightGBM, explains 81.8% of the variance in the engagement score, 2.3 times that of our baseline model. We observed that brand-relative deviation features, which capture how far a post departs from its brand's norms, are the strongest predictors, accounting for 51.3% of total importance. Our research quantifies the extent to which these features are associated with interactions and provides recommendations that organizations may consider.
arXiv:2609.13677v2 公告类型：replace-cross 摘要：现代实际应用问题涉及矩阵值参数，但传统优化器将其视为向量，这促使人们开发利用输入—输出几何结构的矩阵感知方法，例如在参数更新前对动量矩阵进行正交化的 Muon。它在实践中的成功引出了一个概念性问题：正交化动量在非光滑优化中能否依然有效？本文使用与反向传播兼容的广义导数框架，针对局部 Lipschitz 函数研究这一问题。我们的第一项贡献是揭示了一个关键局限：对于任意固定的动量因子 $\beta\in[0,1)$，当步长根据完整的梯度历史自适应调整时，Muon 从几乎所有初始化出发都可能无法趋近某个凸 Lipschitz 目标函数的全局最优解。即使迭代序列有界，也可能出现这种失败。我们的例子受到 Parshakova 等人的例子的启发，而后者仅涵盖 $\beta\in[0,\frac{1}{2})$。随后，我们证明，障碍在于固定动量，而非正交化。事实上，当动量因子自适应调整并趋近于 1，同时步长趋于零时，在有界性和正则性条件下，Muon 可恢复在非凸非光滑优化中的渐近收敛性。此外，我们提出了 MAGD，它将正交化动量与梯度相结合，并根据二者的相对进展进行加权。MAGD 在非凸情形下保持渐近收敛性，并在凸情形下达到 $O(\min\{m,n\}\epsilon^{-2})$ 的收敛速率。一个下界证明了其维度依赖关系的最优性。在合成问题、图像分类和 LLM 预训练上的实验表明，MAGD 是一种简单实用的 Muon 替代方案。总体而言，我们的结果刻画了缺乏光滑性时正交化动量何时会失效，以及如何使其变得可靠；我们希望这些分析能在更广泛的情形下发挥作用。
arXiv:2609.10002v2 公告类型：替换并交叉收录 摘要：深度伪造检测器仍然容易受到基于迁移的黑盒攻击。在此类攻击中，对抗样本在源替代模型上生成，并迁移至攻击者未知的目标模型。然而，源模型与目标模型之间的兼容性如何影响攻击成功率，目前仍缺乏深入理解。既有研究评估的检测器范围有限，且很少区分架构因素与训练因素的影响。我们对涵盖六种骨干网络、两种预训练方案和五种训练数据配置的60个检测器开展了对抗迁移性的受控评估，使用两种攻击方法：AutoAttack（AA）以及结合变换期望的 Carlini--Wagner 攻击（CW--EOT）。匹配比较表明，当源模型与目标模型具有完全相同的骨干网络、相同的架构家族、预训练方案或训练数据时，迁移性显著提高。这种兼容性结构取决于攻击方法：在 AA 下，骨干网络完全一致的影响最大；而在 CW--EOT 下，共享预训练方案和训练数据的影响最大。对非目标源模型的迁移结果取平均时，AA 下的平均攻击成功率（ASR）为 $7.21\%$，CW--EOT 下为 $19.52\%$。相比之下，在排除骨干网络完全一致和训练数据匹配的情况后，结合两种攻击的多源预言机仍达到 \(64.48\%\) 的平均 ASR，表明对源模型取平均可能会大幅低估目标模型的脆弱性。我们公开了240,000张经对抗扰动的图像、完整的两两迁移结果、检测器配置和评估代码。这些发现确立了源模型与目标模型的兼容性以及源模型选择在可信的基于迁移的黑盒鲁棒性评估中的核心地位。
arXiv:2609.04011v2 公告类型：replace-cross 摘要：传递模型是化工过程设计与优化的基础，但其本构定律通常依靠人工选定，实验初始条件被假定为精确值，而计算成本又使优化仅限于少数过程变量。本文提出一种可微混合建模框架，将神经网络嵌入 GPU 加速的群体平衡方程求解器，直接从结晶实验数据中发现本构定律并同化初始条件。利用十三组包含多维尺寸与形状测量结果的真实实验，我们表明，将测得的初始条件视为精确值，尽管能提高预测精度，却会得到物理上不一致的速率定律；而对初始条件进行同化，则能恢复物理上一致的非线性定律，并将轨迹误差降至传统建模方法的七分之一。端到端可微性还支持对高维操作曲线进行基于梯度的优化，找到了一条由 300 个变量描述的温度曲线，在质量不变的条件下将晶体长宽比降低了 25%，从而改善产品属性。我们的框架是对机理建模的增强，而非替代，为从数据中学习传递模型并利用这些模型优化过程提供了一条通用路径。
arXiv:2608.28620v2 公告类型：replace-cross 摘要：偏好获取对于使 AI 系统符合人类价值观至关重要。以往的方法（例如用于器官分配的方法）通常要求利益相关者比较算法的决策（例如选择患者 A 还是患者 B）。这种决策层面的方法混淆了手段与目的。与之不同，我们直接获取对分配结果的偏好，以学习用于策略优化的效用函数。我们构建了一种针对线性效用的新型偏好获取算法，其实际表现优于以往技术。我们的算法分为两个阶段。第一阶段通过成对比较学习切割平面，以快速缩小可能的属性权重空间，并通过排除被支配区域为第二阶段提供热启动。随后，第二阶段以可证明的方式收敛到用户的效用函数。我们将该技术应用于心脏移植分配，这一场景中的策略必须平衡移植后结局、等待名单死亡率、地理便利性和公平性等相互竞争的目标。我们使用该算法开展了一项用户研究，以学习并聚合出符合人类价值观的效用函数。我们将这种线性表示与基于 LLM 的方法进行比较，后者获得了用户提供的自然语言解释，结果发现，这些方法均未比学习得到的线性效用更准确。我们使用得到的效用函数优化心脏移植策略，使其显著更符合人类价值观。与事后最优方案相比，现行策略的竞争比仅为 0.66，而我们的方法以 0.95 的竞争比接近最优。
arXiv:2608.03565v2 Announce Type: replace-cross Abstract: While modern tabular learners excel at capturing statistical patterns, they frequently operate in a semantic vacuum, treating textual features as discrete symbols, ignoring the rich semantics inherent in feature names or cell entries. We propose CASE (Context-Aware Semantic Embeddings), a novel framework that bridges the gap between the semantic understanding of Large Language Models (LLMs) and the statistical capabilities of tabular learners. Unlike existing methods that embed rows in isolation, CASE utilizes a contextualization strategy: we pre-fill the KV cache of a custom-trained Gemma 3-based Tabular Language Model with a representative sample of rows to establish a persistent anchor of the dataset's semantics. This ensures that generated row embeddings are dynamically contextualized, resolving semantic ambiguities and anchoring representations in domain-specific context. Our experiments across several benchmarks (CARTE, TextTab, and TabArena) demonstrate that CASE substantially improves the performance of tabular learners on semantically rich datasets, particularly in low-data regimes.
arXiv:2608.03207v2 公告类型：replace-cross 摘要：pi0 等流匹配视觉-语言-动作（VLA）模型通过对学习得到的去噪速度场进行积分来生成机器人动作，据报道，这类模型能够抵御那些轻易欺骗自回归 VLA 的对抗性扰动。我们表明，这种鲁棒性在很大程度上只是假象：它源于以往的攻击忽略了多步去噪常微分方程（ODE）。我们提出 DRIFT（Denoising Redirection via Input perturbation of the Flow-matching Trajectory），这是一种在测试时放置于机器人夹爪上的通用对抗性贴片，用于攻击现成策略的去噪速度场。我们的核心发现有悖直觉：仅攻击第一个去噪步骤，比攻击更大范围的多个步骤效果更强、成本更低。我们通过输入空间优化所特有的梯度冲突来解释这一现象，而这一现象与训练时的后门攻击情形恰好相反。在四个 LIBERO 测试套件中针对 pi0 和 pi0.5 开展的实验表明，DRIFT 仅用一张小贴片，就能使几乎所有原本可完成的任务失败，其效果远超动作空间和嵌入空间中的攻击基线。
arXiv:2608.02844v2 公告类型：replace-cross 摘要：我们针对梯度难以计算的损失函数，提出了一类基于扩散的通用随机粒子优化方法。具体而言，我们考虑损失梯度可表示为关于参数依赖分布的积分的问题，这类结构涵盖生成模型训练、微调以及潜变量模型学习。我们引入平均场动力学及其相互作用粒子近似，其中包含若干现有算法作为特例，并为构建新方法提供了途径。在适定性和联合收缩性假设下，我们证明了指数收敛性，并表明连续时间粒子系统具有非渐近误差界。我们通过开发高阶 Langevin 优化器和动量优化器，并在最大边际似然估计和基于能量的模型训练任务上对其进行评估，展示了这一框架的应用。
arXiv:2607.21552v2 公告类型：replace-cross 摘要：对于同一个底层问题，采用文本形式还是视觉形式呈现，可能导致视觉语言模型（VLMs）的表现截然不同。这种不一致性表明，不同视角会显露出不同的推理路径和失效模式，而标准的多模态后训练尚未充分利用这些差异。为研究并利用这一现象，我们构建了 ODA-Data，这是一个高质量的配对多模态几何数据集，针对相同问题提供文本主导、图像主导以及图像+文本组合视角，并划分了用于训练和评估视角依赖推理行为的子集。随后，我们提出 Modality-Informed Reciprocal Reasoning Optimization（MIRROR），这是一种强化学习方法，可从模型自身依赖于模态的能力中提取跨视角自蒸馏信号。对于每个问题，MIRROR 会评估模型在所有视角下的表现，选择表现最佳的视角作为教师，并使用 GRPO 任务奖励目标来训练其余视角，同时引入以该教师为目标的反向 KL 正则项。在评估多模态数学问题的各项推理基准上，MIRROR 优于标准强化学习，并使模型在不同输入模态表示下的表现更加准确、一致。
arXiv:2607.20274v3 公告类型：替换并交叉列出 摘要：诊断分类器和影像生物标志物基于医学基础模型的嵌入表示进行拟合。随着新版本出现，这些模型会被替换。这种做法假定不同模型对图像的表示方式相似。我们使用来自5种成像模态、14个数据集的超过750,000张图像、18个公开模型，以及101个在胸部X线片和组织病理图像上训练的模型，检验了这一假设。这101个模型在预训练目标、标签类型、规模、随机种子、初始化、训练所用患者或训练方案方面存在差异。我们以相互k近邻重叠率衡量一致性，即一张图像在两个模型中的最近邻集合共有元素所占的比例。对于一张胸部X线片，两个公开模型的10个最近邻平均重叠比例为0.091。随机初始化的网络与这些模型的重叠比例为0.036。在其余四种模态中，两个公开模型的重叠比例最高为0.300。在胸部X线片上，一致性更多地取决于来源数据集和摄片体位，而非影像所见，且与准确率无显著关联。利用4,096张无标签X线片拟合两个模型嵌入表示之间的线性映射后，可迁移针对14种胸部影像所见的分类器，其受试者工作特征曲线下面积达到原始值的0.987。趋同性取决于预训练目标。在胸部X线摄影和组织病理学中，标签监督模型在两种模型规模下的趋同性均最低。在胸部X线片上，仅随机种子不同的两个标签监督模型，其最近邻重叠比例为0.062至0.100。而采用自蒸馏、掩码图像建模或对比预训练的两个模型，其最近邻重叠比例为0.411至0.802。因此，开发者在选择预训练目标时，也可以选择模型的趋同性。在替换模型之前，用户可以使用无标签图像检验后继模型是否返回相同的最近邻。
arXiv:2607.17679v2 Announce Type: replace-cross Abstract: This is an invited commentary on the Psychometrika focus article "Fairness Issues and Evaluation in Psychometrics and AI/ML: What Can We Learn from Each Field?" by Ying Cheng (2026, doi:10.1017/psy.2026.10110). Cheng offers a systematic comparison between long-standing test fairness and modern algorithmic fairness. Her mapping of the entire testing workflow onto the AI/ML fairness paradigm, rather than only the final selection stage, is a crucial contribution to interdisciplinary fairness research. This commentary extends her discussion by examining two conceptual issues: the distinction between equality and equity, and the role of causality in fairness research. The focus article and this commentary point to directions for future fairness research across the psychometrics and AI/ML communities.
arXiv:2607.02592v3 Announce Type: replace-cross Abstract: On-policy distillation (OPD) has recently emerged as an effective post-training paradigm by providing supervision on student-generated trajectories. However, existing OPD methods for multimodal reasoning usually rely on a static teacher routing, assigning each sample to a single teacher based on modality or task type. This ignores that visual grounding and abstract reasoning may dominate different decoding steps, making a single teacher insufficient for the full trajectory. To this end, H-OPD is proposed as a confidence-aware heterogeneous multi-teacher OPD framework for multimodal reasoning. By verifying the complementarity of heterogeneous teachers in the same reasoning process, H-OPD replaces task or sample level teacher routing with token-level teacher arbitration along the shared student trajectory. H-OPD employs vision-to-language description transfer to enable text-only teachers to access key visual semantics, and uses a confidence-aware arbitration mechanism to dynamically combine vision-language teacher and text-only teachers at each token. Extensive evaluations over 11 widely-used reasoning benchmarks showcase the superior performance of our method.