知识热榜

62 个榜单
共 50 条热搜查看完整榜单
  • 01
    Young organs may not be a fountain of youth for recipients
    Around this time last year I was attending an aging conference in Manchester, listening to a talk about fly aging, when my phone started pinging. News outlets were reporting that a hot mic had caught Russia’s and China’s leaders discussing the possibility of living forever. “With the developments of biotechnology, human organs can be continuously…Jessica Hamzelou
  • 02
    The specter of AI-enabled bioweapons is a wake-up call for biotech
    In recent weeks, leaders of some of the biggest AI companies have warned that the very tech they are developing is dangerous. Last weekend, Anthropic CEO Dario Amodei argued that AI carries serious risk and that progress should be slowed. OpenAI CEO Sam Altman responded on X: “I agree with Dario that we need to pace the…Jessica Hamzelou
  • 03
    Meet a mouse whose brain cortex is made up of human cells
    Multiple cameras tracked a mouse as it wandered around a small arena. A computer charted its position and speed, leaving Pong-like traces on a monitor. The reason to watch this rodent so carefully? Nearly half its brain volume had been replaced with human cells. The effort to mix the brain tissues of distant species is…Antonio Regalado
  • 04
    AI models need more data about biology, and OpenAI is paying to create it
    Last year Ruxandra Teslo, a policy analyst who focuses on clinical trials, posted an idea for supercharging medical AI systems: Use data from failed biotech companies. By bidding at their bankruptcy proceedings, she proposed, it might be possible to obtain detailed regulatory filings, manufacturing strategies, and safety data—types of information usually considered trade secrets. She…Antonio Regalado
  • 05
    Donated livers can be made biologically younger
    Once an organ is removed from a donor’s body, the clock starts ticking. Surgeons usually flush the organ with a preservative solution, bag it, and put it on ice—where it immediately starts to degrade. The team has a matter of hours to get it into a recipient’s body. There’s another option—one that has been growing…Jessica Hamzelou
  • 06
    Meet the under-35s shaping the future of biotech
    Every year, MIT Technology Review puts together a list of some of the brightest and best young minds working across science and technology. Our 35 Innovators Under 35 are the ones to watch—people whose research and technical work stands to shape the future of their fields. This year, the list includes nine people who are…Jessica Hamzelou
  • 07
    Healthcare AI’s next test is integration
    The entrance of major AI companies into healthcare is a meaningful and welcome development, accelerating the technical foundation available to the industry. Their models are increasingly capable of processing long clinical records, interpreting complex terminology, comparing documentation against evidence and generating coherent summaries from large volumes of information. For clinicians, operators, and administrative teams who…Andrew Ray
  • 08
    This geneticist’s age-reversal tech could help restore sight
    Yuancheng (Ryan) Lu is obsessed with aging. And with eyes. As he steps outside the Whitehead Institute in Cambridge, Massachusetts, his aviator glasses darken automatically in the sun. Age-related blindness runs in his family. A great-aunt in China, the story goes, was killed crossing a road because she couldn’t see oncoming traffic. And Lu’s own…Antonio Regalado
  • 09
    A startup claims it’s found a drug to make your blood young
    I knew I’d officially become a “longevity influencer” this month when a company called Generation Lab reached out to offer me the chance to write about—and even receive—their new rejuvenation treatment, an injectable combination of two existing drugs which they call 1 Generation. This wasn’t just any antiaging treatment, either. A company fact sheet says that…Antonio Regalado
  • 10
    Welcome to the spiderverse, a world measured through webs
    Counting the creatures in the world around us is critical for a raft of conservation efforts. It helps scientists gauge biodiversity, track migration, and spot invasive species. That census-taking, though, often requires humans to tabulate what they see, trap, or otherwise sense—a potentially laborious, costly process that can still leave gaps. But developments over the…Stephen Ornes
共 10 条热搜查看完整榜单

Nature · 实时热榜

5小时前更新
共 75 条热搜查看完整榜单
共 100 条热搜查看完整榜单

OEIS · 近期新增

18分钟前更新
共 10 条热搜查看完整榜单

OpenAlex · 实时热榜

14分钟前更新
共 50 条热搜查看完整榜单
  • 01
    DecompRL: Solving Harder Problems by Learning Modular Code Generation
    期刊：Transactions on Machine Learning Research · 摘要：How can Large Language Models (LLMs) solve new problems? Repeated sampling raises the chance of success by scaling test-time compute, but its GPU cost grows linearly with the number of attempts, while reinforcement learning (RL) with verifiable rewards improves single-attempt accuracy at the expense of sample diversity. Both strategies ultimately fail when the base policy has near-zero probability of producing a correct solution: no amount of sampling or gradient signal can overcome a search space that is simply too large. We take a different approach: rather than sampling harder, we make the… · 篇幅：Regular submission (no more than 12 pages of main content) · OpenReview ID：RnBSdiMyBZJuliette Decugis, Fabian Gloeckle, Francis Bach et al.
  • 02
    Evaluating the evidence trace of NeurIPS 2025 contributions with agentic code auditing
    期刊：Transactions on Machine Learning Research · 摘要：Code serves as the primary evidence behind computational publications, yet a detailed review of an unfamiliar codebase imposes a commonly prohibitive time burden on volunteer reviewers. Consequently, self-reported reproducibility checklists at major machine learning venues are rarely empirically verified. This leaves code as a significant blind spot in peer review. To bridge this gap, we introduce AuditOwl, an autonomous, verification-centric LLM pipeline designed to make code auditing feasible for authors pre-submission and reviewers post-submission. With this framework, we conduct an audit… · 篇幅：Regular submission (no more than 12 pages of main content) · 代码：https://github.com/PascalIversen/auditowl-neurips2025 · OpenReview ID：xjCJY8XxcmPascal Iversen, Ferdous Nasri, Bernhard Y Renard et al.
  • 03
    ShapeY: A Principled Framework for Measuring Shape Recognition Capacity via Nearest-Neighbor Matching
    期刊：Transactions on Machine Learning Research · 摘要：In humans, the ability to recognize and categorize objects despite variations in 3D viewpoint, lighting, surface properties, optical conditions, etc., is known to depend primarily on shape. In contrast, state-of-the-art artificial vision systems based on deep networks rely on an intricate mixture of shape, texture and color cues that defies simple characterization. To better understand the capabilities of machine vision systems specifically with regard to shape, we developed ShapeY, a novel recognition benchmark that uses nearest-neighbor matching of images containing only shape information t… · 篇幅：Long submission (more than 12 pages of main content) · 代码：https://github.com/njw0709/ShapeY · OpenReview ID：cTm6EecEBYJong Woo Nam, Amanda Sofie Rios, Bartlett W Mel
  • 04
    Where Not to Learn: Prior-Aligned Training with Subset-based Attribution Constraints
    期刊：Transactions on Machine Learning Research · 摘要：Reliable models should not only predict correctly, but also base their decisions on acceptable evidence. However, conventional supervised learning typically provides only class-level labels, allowing models to achieve high accuracy by exploiting shortcut correlations rather than intended decision evidence. Human priors, such as bounding boxes or target interface elements, can help constrain such behavior, but aligning model evidence with these priors remains challenging because learned decision evidence often diverges from human perception. In this work, we study attribution–prior alignment w… · 篇幅：Regular submission (no more than 12 pages of main content) · 代码：https://github.com/RuoyuChen10/Attribution_Constraints · OpenReview ID：2katjvBbICRuoyu Chen, Shangquan Sun, Xiaoqing Guo et al.
  • 05
    How Much Quality Survives LLM Downshifting? Judge Disagreement, Protocol Sensitivity, and a Matched Benchmark–Traffic Difference of at Most Three Points
    期刊：Transactions on Machine Learning Research · 摘要：Teams cut large language model (LLM) inference cost by routing each request to the smallest model that is good enough, instead of always calling a single frontier model. To justify the switch they quote one headline number: “the small model preserves about X% of quality.” We show that this headline cannot be trusted. We run the most aggressive policy possible, a static 70B→8B downshift within one open-weight family, over a 10,000-prompt curated benchmark spanning nine task types, and make four findings. 1. The cost reduction is large and judge-independent but price-assumption-dependent: token… · 篇幅：Regular submission (no more than 12 pages of main content) · OpenReview ID：6ccINAFoJiNarasingaMoorthy VeiluKanthaPerumal
  • 06
    A Controlled Synthetic Benchmark for Educational Aspect-Based Sentiment Analysis
    期刊：Transactions on Machine Learning Research · 摘要：Large language models are increasingly used to synthesize labeled training data where annotation is scarce, but the generated labels are often unfaithful to the text, and it is difficult to tell whether a synthetic benchmark carries genuine learnable signal or merely reproduces label frequencies. We study both problems in educational aspect-based sentiment analysis (ABSA), a setting where real aspect-labeled feedback is private and costly to annotate, with a methodology that applies beyond it. We release a controlled corpus of 10,000 synthetic course reviews over a 20-aspect pedagogical schem… · 篇幅：Long submission (more than 12 pages of main content) · OpenReview ID：tmNAKpmSaFYehudit Aperstein, Alexander Apartsin
  • 07
    Active Sampling for Ultra-Low-Bit-Rate Video Compression via Conditional Controlled Diffusion
    期刊：Transactions on Machine Learning Research · 摘要：Diffusion models provide a powerful generative prior for perceptual reconstruction at ultra-low bitrates, but effective video compression requires controlling the generative process using highly compact conditioning signals. In this work, we present ActDiff-VC, a diffusion-based video compression framework for the ultra-low-bitrate regime. Our method partitions videos into variable-length segments, transmits keyframes only when needed, and summarizes temporal dynamics using a compact set of tracked point trajectories. Conditioned on these sparse signals, a conditional diffusion decoder synthe… · 篇幅：Regular submission (no more than 12 pages of main content) · 代码：https://github.com/importAmir/ActDiff-VC · OpenReview ID：zER4e4GfZHAmirhosein Javadi, Shirin Saeedi Bidokhti, Tara Javidi
  • 08
    Mechanistic Interpretability of Transformer MLPs Through Exact Soft‑Gate Decomposition
    期刊：Transactions on Machine Learning Research · 摘要：Transformer feed-forward (MLP) blocks perform nonlinear transformations, but their internal structure is often difficult to interpret using current methods. This paper presents an exact soft-gate decomposition that extracts the effective weight matrix for each MLP block at varying inputs. This approach generalises exact piecewise-affine decomposition from ReLU networks to Gaussian Error Linear Unit (GELU) and Sigmoid Linear Unit (SiLU) activations, covering both standard and gated MLP architectures. Verification across five models, including BERT-base, mBART-50, and GPT-2 Small for language t… · 篇幅：Long submission (more than 12 pages of main content) · 代码：https://doi.org/10.5281/zenodo.22959811 · OpenReview ID：20lTANRDAzArnab Barua, Mobyen Uddin Ahmed, Shahina Begum
  • 09
    CountLoop: Training-Free High-Instance Image Generation via Iterative Agent Guidance
    期刊：Transactions on Machine Learning Research · 摘要：Diffusion models excel at photorealistic synthesis but struggle with object count fidelity, especially in high-density settings. We introduce COUNTLOOP, a training-free framework that achieves structured instance and count control through iterative, structured feedback. Our method alternates between synthesis and evaluation: a VLM-based planner generates structured scene layouts, while a VLM-based critic provides explicit feedback on object counts, spatial arrangements, and visual quality to refine the layout iteratively. Instance-driven attention masking and cumulative attention composition… · 篇幅：Regular submission (no more than 12 pages of main content) · 代码：https://github.com/mondalanindya/CountLoop/ · OpenReview ID：2JxXGhpCP4Anindya Mondal, Sauradip Nag, Ayan Banerjee et al.
  • 10
    LEAD: An EEG Foundation Model for Alzheimer’s Disease Detection
    期刊：Transactions on Machine Learning Research · 摘要：Electroencephalography (EEG) provides a non-invasive, highly accessible, and cost-effective approach for detecting Alzheimer’s disease (AD). However, existing methods, whether based on handcrafted feature engineering or standard deep learning, face three major challenges: 1) the lack of large-scale EEG-based AD datasets for robust representation learning and evaluation; 2) limited cross-subject generalizability; and 3) difficulty in adapting to highly heterogeneous data. To address these challenges, we curate the world’s largest EEG-AD corpus to date, comprising 2,238 subjects. Leveraging thi… · 篇幅：Long submission (more than 12 pages of main content) · 代码：https://github.com/DL4mHealth/LEAD · OpenReview ID：AigNTyxcvHYihe Wang, Nan Huang, Nadia Mammone et al.
  • 11
    Training-Free Modality-Agnostic Concept Sliders: Fine-Grained Control via Diffusion Models of Images, Audio, and Video
    期刊：Transactions on Machine Learning Research · 摘要：Diffusion models have become state-of-the-art generative models for images, audio, and video, yet enabling fine-grained controllable generation, i.e., continuously steering specific concepts without disturbing unrelated content, remains challenging. Concept Sliders (CS) offer a promising direction by discovering semantic directions through textual contrasts, but they require per-concept training and architecture-specific fine-tuning (e.g., LoRA), limiting scalability to new modalities. In this work, we introduce a simple yet effective approach that is fully training-free and modality-agnostic… · 篇幅：Regular submission (no more than 12 pages of main content) · 代码：https://github.com/azencot-group/Free_Sliders · OpenReview ID：m2hpORnHhFRotem Ezra, Hedi Zisling, Nimrod Berman et al.
  • 12
    Neural Operators for Multi-Task Control and Adaptation
    期刊：Transactions on Machine Learning Research · 摘要：Neural operator methods have emerged as powerful tools for learning mappings between infinite-dimensional function spaces, yet their potential in optimal control remains largely unexplored. We focus on multi-task control problems, whose solution is a mapping from task description (e.g., cost or dynamics functions) to optimal control law (e.g., feedback policy). We approximate these solution operators using a permutation-invariant neural operator architecture. Across a range of parametric optimal control environments and a locomotion benchmark, a single operator trained via behavioral cloning… · 篇幅：Long submission (more than 12 pages of main content) · 代码：https://github.com/CLeARoboticsLab/neural-operator-control · OpenReview ID：jciOb0z5WmDavid SeWell, Xingjian Li, Stepan Tretiakov et al.
  • 13
    Tackling Decision Processes with Non-Cumulative Objectives using Reinforcement Learning
    期刊：Transactions on Machine Learning Research · 摘要：Markov decision processes (MDPs) are used to model a wide variety of applications ranging from game playing over robotics to finance. Their optimal policy typically maximizes the expected sum of rewards given at each step of the decision process. However, many real-world problems do not fit straightforwardly into this framework: Non-cumulative Markov decision processes (NCMDPs), where instead of the expected sum of rewards, the expected value of an arbitrary function of the rewards is maximized. Example functions include the maximum of the rewards or their mean divided by their standard devia… · 篇幅：Regular submission (no more than 12 pages of main content) · 代码：https://github.com/MaxNaeg/ncmdp · OpenReview ID：OJCdddDtNyMaximilian Nägele, Jan Olle, Thomas Fösel et al.
  • 14
    RMB: Reward Model Boosting Mitigates Reward Hacking
    期刊：Transactions on Machine Learning Research · 摘要：Reinforcement Learning from Human Feedback (RLHF) is a powerful technique for aligning large language models (LLMs) with human preference. However, it often suffers from the reward hacking issue, where policy optimization improves the proxy reward model while actually degrading performance with respect to the true human preference, due to the imperfection of the proxy. To address this, we propose Reward Model Boosting (RMB), a novel approach that enhances the robustness and reliability of the reward signal for RLHF. RMB first trains a set of reward models with a diverse-promoting regularizer.… · 篇幅：Regular submission (no more than 12 pages of main content) · 代码：https://github.com/MANGA-UOFA/RMB · OpenReview ID：bIaCQSyb4gJiabin Fan, Dezhi Ye, Yongchang Hao et al.
  • 15
    INDEQS: Informed Neural controlled Differential EQuationS
    期刊：Transactions on Machine Learning Research · 摘要：Neural Controlled Differential Equations (NCDE) provide a powerful continuous-time framework for forecasting time series, but standard graph-based extensions typically learn spatial structure purely from data, even in settings where a directed graph structure is known a priori. We introduce Informed Neural controlled Differential EQuationS (INDEQS), a modification to graph-based NCDE forecasting methods that incorporates prior knowledge of a directed graph at distinct architectural positions. INDEQS separates inner mixing of hidden states across graph nodes from outer mixing between vector fi… · 篇幅：Long submission (more than 12 pages of main content) · 代码：https://github.com/mitchi1/indeqs · OpenReview ID：okGwJeKlZ4Michael Detzel, Gabriel Nobis, Kristiyan Blagov et al.
  • 16
    FedPS: Federated Preprocessing for structured data via aggregated Statistics
    期刊：Transactions on Machine Learning Research · 摘要：Federated Learning (FL) enables multiple parties to collaboratively train machine learning models without sharing raw data. However, before training, data must be preprocessed to address missing values, inconsistent formats, and heterogeneous feature scales. This preprocessing stage is critical for model performance but is largely overlooked in FL research. In practical FL systems, privacy constraints prohibit centralizing raw data, while communication efficiency introduces further challenges for distributed preprocessing. We introduce FedPS, a framework for federated data preprocessing based… · 篇幅：Long submission (more than 12 pages of main content) · 代码：https://github.com/xuefeng-xu/fedps · OpenReview ID：MdeXZVjNKuXuefeng Xu, Graham Cormode
  • 17
    BTNR: Bayesian Tensor Networks for Regression
    期刊：Transactions on Machine Learning Research · 摘要：Universal function approximation via basis expansions such as Fourier or polynomial bases is resource intensive. While the number of coefficients grows moderately in low dimensions, it can scale exponentially with the input size, leading to high complexity and potential overfitting. Tensor networks compress the coefficient space allowing for linear rather than exponential scaling. Existing tensor networks methods derive their learning equations for one specific network topology and few studies explore Bayesian learning methods. Probabilistic frameworks naturally derive confidence estimates fo… · 篇幅：Regular submission (no more than 12 pages of main content) · 代码：https://git.kosmon.org/nicco/BTNR · OpenReview ID：5yPMm3hq7kNiccolò Ciolli, Jesper Løve Hinrich, Morten Mørup
  • 18
    OpInf-LLM: Parametric PDE Solving with LLMs via Operator Inference
    期刊：Transactions on Machine Learning Research · 摘要：Solving diverse partial differential equations (PDEs) is fundamental in science and engineering. Large language models (LLMs) have demonstrated strong capabilities in code generation, symbolic reasoning, and tool use, but reliably solving PDEs across heterogeneous settings remains challenging. Prior work on LLM-based code generation and transformer-based foundation models for PDE learning has shown promising advances. However, a persistent trade-off between execution success rate and numerical accuracy arises, particularly when generalization to unseen parameters and boundary conditions is re… · 篇幅：Regular submission (no more than 12 pages of main content) · 代码：https://github.com/jacobwang925/opinf-llm · OpenReview ID：rCTRgstMr1Zhuoyuan Wang, Hanjiang Hu, Xiyu Deng et al.
  • 19
    A Comprehensive Survey on 3D Deep Point Cloud Models
    期刊：Transactions on Machine Learning Research · 摘要：Recently, point cloud data has attracted the attention of researchers as a promising data representation model for a wide range of applications. As unlike 2D data, point clouds are unordered, irregular, and often large in scale, they might impose severe challenges when designing deep learning models. Over the past decade, substantial progress has been made in proposing architectures that address permutation invariance, geometric reasoning, scalability, and robustness, leading to rapid expansion across diverse 3D data oriented applications. The main aims of this paper are to present a comprehe… · 篇幅：Long submission (more than 12 pages of main content) · 代码：https://github.com/IPL-Sharif/3D_Survey · OpenReview ID：WpQdfOC36sZeynab Yasamani, Amir M. Mansourian, Parniya Seifi et al.
  • 20
    Generate, Filter, Control, Replay: A Comprehensive Survey of Rollout Strategies for LLM Reinforcement Learning
    期刊：Transactions on Machine Learning Research · 摘要：Reinforcement learning (RL) has become a central post-training tool for improving the reasoning abilities of large language models (LLMs). In these systems, the rollout, the trajectory sampled from a prompt to termination, including intermediate reasoning steps and optional tool or environment interactions, determines the data that the optimizer ultimately learns from, yet rollout design is often treated as an implementation detail and underreported. This survey provides an optimizer-agnostic view of rollout strategies for RL-based post-training of reasoning LLMs. We formalize rollout pipelin… · 篇幅：Long submission (more than 12 pages of main content) · OpenReview ID：aB848FutiRRohan Surana, Gagan Mundada, Xunyi Jiang et al.
  • 21
    Post-Anomaly Detection Inference for Deep SVDD
    期刊：Transactions on Machine Learning Research · 摘要：Deep Support Vector Data Description (Deep SVDD) has become a prominent framework for unsupervised anomaly detection by learning latent representations that compactly characterize normal data around a center. Despite its empirical success, anomaly decisions produced by Deep SVDD are typically made solely based on anomaly scores without rigorous statistical guarantees, thereby limiting their reliability in safety-critical and high-stakes applications where false positives must be strictly controlled. In this paper, we propose PADI (Post-Anomaly Detection Inference), a novel framework that equi… · 篇幅：Long submission (more than 12 pages of main content) · 代码：https://github.com/DAIR-Group/PADI · OpenReview ID：f8XTHjxBigCao Le Cong Thanh, Dang Quang Vinh, Vo Nguyen Le Duy
  • 22
    When Should the Agent Speak? A Survey of Intervention Timing for Always-On AI Assistants
    期刊：Transactions on Machine Learning Research · 摘要：Agents now complete multi-step tasks against verifiable goals, and egocentric perception resolves user intent with high reported accuracy. What remains unmodeled is the decision of when to act unprompted. We survey intervention timing for always-on assistants (smart glasses, MR headsets, ambient copilots) around a single decision rule, intervene iff the expected benefit of acting exceeds the expected cost of interrupting, and organize the literature into five layers: signals, decision, action, memory, and evaluation. We reconnect two lineages that have proceeded almost without citing each oth… · 篇幅：Long submission (more than 12 pages of main content) · 代码：https://github.com/tao-hpu/awesome-proactive-agents · OpenReview ID：b0yKEdAXErTao An
  • 23
    Cross-Layer Discrete Concept Discovery for Interpreting Language Models
    期刊：Transactions on Machine Learning Research · 摘要：Interpreting language models remains challenging due to the residual stream, which linearly mixes and duplicates features across adjacent layers, causing single-layer analyses to miss this cross-layer structure. Cross-layer sparse autoencoders (SAEs) address layer mixing but operate in continuous space, where concepts split across many neurons without clear boundaries. We introduce cross-layer vector quantized-variational autoencoder (CLVQ-VAE), a novel framework that maps representations from a lower layer to a higher layer through a discrete vector-quantization bottleneck, collapsing duplic… · 篇幅：Regular submission (no more than 12 pages of main content) · 代码：https://github.com/agarg-dev/CLVQVAE/tree/main · OpenReview ID：XbBvGKSoEGAnkur Garg, Xuemin Yu, Hassan Sajjad et al.
  • 24
    A Journey Through Surface Convolutions
    期刊：Transactions on Machine Learning Research · 摘要：Surface data appears in a wide range of machine learning tasks. Among other objectives, geometric deep learning generalizes Euclidean CNNs to two-dimensional Riemannian manifolds, allowing to learn patterns of data along curved surfaces. While the literature presents numerous surface convolutions, existing comparisons often vary both the convolution and the underlying surface charting algorithm, complicating the isolation of the source of performance differences. Furthermore, unlike in Euclidean CNNs, the size of the convolution neighborhood is not implied by the template size but manually se… · 篇幅：Long submission (more than 12 pages of main content) · 代码：https://github.com/andreasMazur/geoconv · OpenReview ID：lCwv0bo973Andreas Mazur, David Philip Leins, Fabian Hinder et al.
  • 25
    Towards Bridging the Gap Between Offline and Iterative Alignment via Preference Distillation
    期刊：Transactions on Machine Learning Research · 摘要：Direct preference optimization (DPO) is a promising offline approach for aligning large language models (LLMs) due to its simplicity, computational efficiency, and implicit modeling of human preferences. Interestingly, iterative extensions of DPO have achieved stronger performance on academic benchmarks, raising two key questions: (i) Why do iterative methods generally outperform offline ones? (ii) Can their advantages be incorporated into offline alignment? To answer the first question, our controlled experiments reveal that the \textit{explicit preference model}, additionally introduced in… · 篇幅：Regular submission (no more than 12 pages of main content) · 代码：https://github.com/onepounchman/DP3O · OpenReview ID：X6r1bU1m6xWenbo Zhang, Wenzhuo Zhou, Hengrui Cai et al.
  • 26
    Searching for Actual Causes: Approximate Algorithms with Adjustable Precision
    期刊：Transactions on Machine Learning Research · 摘要：Causality has gained increasing attention in recent years, notably for improving the interpretability of machine learning models. Yet the field of explainable artificial intelligence (XAI) has been criticized for emphasizing general tendencies rather than the situation-specific facts, which users typically expect as explanations. These expectations align with the notion of actual causes, which identify what made the observed outcome happen, in the specific context at hand. Halpern and Pearl provided a formal basis for actual causation, but identifying actual causes is NP-complete. Practical i… · 篇幅：Long submission (more than 12 pages of main content) · 代码：https://github.com/SamuelReyd/SearchingForCauses · OpenReview ID：3NlYZPCH9vSamuel Reyd, Ada Diaconescu, Jean-Louis Dessalles
  • 27
    Wavenumber-Resolved Spectral Gating for Diffusion Models of Two-Dimensional Turbulence
    期刊：Transactions on Machine Learning Research · 摘要：We study unconditional generative modeling of two-dimensional Kolmogorov flow with denoising diffusion. A plain U-Net diffusion model reproduces the point-wise statistics of turbulent vorticity fields but distorts the inertial-range spectrum and the spectral fluxes that carry energy and enstrophy across scales. We add a wavenumber-resolved spectral gate: a small Fourier-domain bottleneck that learns a per-channel multiplicative correction over a coarse radial-wavenumber grid, conditioned on the diffusion noise level and a continuous log-viscosity parameter. The gate is paired with physics los… · 篇幅：Long submission (more than 12 pages of main content) · 代码：https://github.com/nishattasnim15215/Wavenumber-Resolved-Spectral-Gating-for-Diffusion-Models-of-Two-Dimensional-Turbulence · OpenReview ID：MhYd4okrNaNishat Tasnim, Yong Shi
  • 28
    Automated Membership Inference Attacks (AutoMIA): Discovering MIA Signal Computations using LLM Agents
    期刊：Transactions on Machine Learning Research · 摘要：Membership inference attacks (MIAs), which enable adversaries to determine whether specific data points were part of a model's training dataset, have emerged as an important framework to understand, assess, and quantify the potential information leakage associated with machine learning systems. Designing effective MIAs is a challenging task that usually requires extensive manual exploration of model behaviors to identify potential vulnerabilities. In this paper, we introduce AutoMIA -- a novel framework that leverages large language model (LLM) agents to automate the design and implementation… · 篇幅：Regular submission (no more than 12 pages of main content) · OpenReview ID：N3VOIYIqo9Toan Tran, Olivera Kotevska, Li Xiong
  • 29
    Beyond Neural Collapse: Task-Intrinsic Geometry Governs Neural Representations in Modular Arithmetic
    期刊：Transactions on Machine Learning Research · 摘要：While neural collapse (NC) predicts that a $K$-class-balanced classifier should organize terminal representations as a $(K-1)$-dimensional simplex equiangular tight frame (ETF), modular addition consistently enters a different regime: networks compress to a two-dimensional cyclic geometry in which both classifier weights and token embeddings lie on circles. We refine the explanation of this phenomenon in three directions. First, we formalize a layerwise non-uniform training mechanism: downstream classifier weights are driven by dense cross-entropy gradients into a rank-2 equiangular configura… · 篇幅：Regular submission (no more than 12 pages of main content) · OpenReview ID：ve7Iq6JQ6GHu Tan, Kuo Gai, Shihua Zhang
  • 30
    Inducing Artificial Uncertainty in Language Models
    期刊：Transactions on Machine Learning Research · 摘要：In safety-critical applications, language models should be able to characterize their uncertainty with meaningful probabilities. Many uncertainty quantification approaches require supervised data; however, finding suitable unseen challenging data is increasingly difficult for large language models trained on vast amounts of scraped data. If the model is consistently (and correctly) confident in its predictions, the uncertainty quantification method may consistently overestimate confidence on new and unfamiliar data. Finding data which exhibits enough uncertainty to train supervised uncertaint… · 篇幅：Regular submission (no more than 12 pages of main content) · 代码：https://github.com/sophia-hager/artificial-uncertainty/tree/main · OpenReview ID：cUcWgu0wGUSophia Hager, Simon Zeng, Nicholas Andrews
  • 31
    Implicit Bias and Invariance: How Hopfield Networks Efficiently Learn Graph Orbits
    期刊：Transactions on Machine Learning Research · 摘要：Many learning problems are organized by group symmetries. While invariance is often imposed through architectures or group averaging, we ask when it can emerge from training on a finite random subset of an orbit. We study this question in classical Hopfield networks, where strict memorization can be expressed as a linear margin problem. Reparameterizing minimization of energy flow (MEF) as an exponential loss connects gradient descent to the corresponding minimum-norm hard-margin memorizer. Our main result shows that, for independent uniform samples from any finite permutation orbit, the exac… · 篇幅：Regular submission (no more than 12 pages of main content) · OpenReview ID：Fptb4mpPS2Michael Murray, Tenzin Chan, Kedar Karhadkar et al.
  • 32
    MeritKV: Utility-Gated Admission Control for KV-Cache Reuse in LLM Serving
    期刊：Transactions on Machine Learning Research · 摘要：Key-value (KV) cache reuse lowers prefill latency, but lookup, cache manipulation, and speculative precompute can cost more than the computation they save, so a serving system can be slower than no cache even at a high hit rate. We argue that KV reuse should be a per-request economic decision rather than an unconditional mode, and present MeritKV, a lightweight admission-control layer that scores each candidate reuse by its net utility (saved prefill time minus reuse cost minus expected waste) and executes it only when that estimate is positive, bypassing otherwise. The gate is backend-agnost… · 篇幅：Regular submission (no more than 12 pages of main content) · 代码：https://github.com/Kushalk0677/meritkv · OpenReview ID：y3j6fucWQPKushal Khemani, Evan J. Leri, Sparsh Mittal
  • 33
    Inverse Problems Conditioned on Observation Ensembles: Applications and Methods
    期刊：Transactions on Machine Learning Research · 摘要：We introduce a new multivariate statistical problem that we refer to as the Ensemble-conditioned Inverse Problem (EIP). The aim of EIP is to invert for an ensemble that is distributed according to the pushforward of a prior under a forward process. In high energy physics (HEP), this is related to a widely known problem called unfolding, which aims to reconstruct the true physics distribution from observations that are distorted by detector effects. The EIP also arises in full waveform inversion (FWI) and inverse imaging with unknown priors. We propose non-iterative inference-time methods that… · 篇幅：Regular submission (no more than 12 pages of main content) · 代码：https://github.com/ZhengyanHuan/EIP · OpenReview ID：XG3BcxSex0Zhengyan Huan, Camila Pazos, Martin Klassen et al.
  • 34
    Can LLMs Grasp Implicit Cultural Values? Benchmarking LLMs' Cultural Intelligence with CQBench
    期刊：Transactions on Machine Learning Research · 摘要：Cultural Intelligence (CQ) refers to the ability to understand unfamiliar cultural contexts—a crucial skill for large language models (LLMs) to effectively engage with globally diverse users. Existing studies often focus on explicitly stated cultural norms, but fail to capture the subtle, implicit values that are common in daily conversation. To address this gap, we introduce CQ-Bench, a benchmark specifically designed to assess LLMs’ capability to infer implicit cultural values from natural conversational contexts. CQ-Bench consists of multi-character conversation-based stories using values… · 篇幅：Long submission (more than 12 pages of main content) · OpenReview ID：eIdvFTtuymZiyi Liu, Priyanka Dey, Jen-tse Huang et al.
  • 35
    Do Instance Priors Help Weakly Supervised Semantic Segmentation?
    期刊：Transactions on Machine Learning Research · 摘要：Semantic segmentation requires dense pixel-level annotations, which are costly and time-consuming to acquire. To address this, we present SeSAM, a framework that uses a foundational segmentation model, i.e. Segment Anything Model (SAM), with weak labels, including coarse masks, scribbles, and points. SAM, originally designed for instance-based segmentation, cannot be directly used for semantic segmentation tasks. In this work, we identify specific challenges faced by SAM and determine appropriate components to adapt it for class-based segmentation using weak labels. Specifically, SeSAM decomp… · 篇幅：Regular submission (no more than 12 pages of main content) · OpenReview ID：4bHFNKe8OUAnurag Das, Anna Kukleva, Xinting Hu et al.
  • 36
    Structural Bias Beyond Homophily: A Study of Fairness in Link Prediction
    期刊：Transactions on Machine Learning Research · 摘要：Graph link prediction (LP) plays a critical role in socially impactful applications such as job recommendation and friendship formation, making fairness an essential concern in this task. While many fairness-aware methods manipulate graph structures to mitigate prediction disparities, the topological biases inherent to social graphs remain poorly understood and are consistently conflated with homophily alone. In this work, we study the relationship between structural biases and fairness outcomes in LP. To this end, we formalize a taxonomy of topological bias measures and introduce a graph gen… · 篇幅：Regular submission (no more than 12 pages of main content) · 代码：https://github.com/lilianmarey/Structural_Bias_TMLR · OpenReview ID：xMQ5v0cxH3Lilian Marey, Mathilde Perez, Tiphaine Viard et al.
  • 37
    Function Space Diversity for Uncertainty Prediction via Repulsive Last-Layer Ensembles
    期刊：Transactions on Machine Learning Research · 摘要：Bayesian inference in function space has gained attention due to its robustness against overparameterization in neural networks. However, approximating the infinite-dimensional function space introduces several challenges. In this work, we discuss function space inference via particle optimization and present practical modifications that improve uncertainty estimation and, most importantly, make it applicable for large and pretrained networks. First, we demonstrate that the input samples, where particle predictions are enforced to be diverse, are detrimental to the model performance. While di… · 篇幅：Regular submission (no more than 12 pages of main content) · 代码：https://github.com/stegsoph/function-space-last-layer-ensembles · OpenReview ID：G8fAj5gzEpSophie Steger, Christian Knoll, Bernhard Klein et al.
  • 38
    Scale-Invariant, Robust Sparse PCA for Large Data via Differentiable Penalties
    期刊：Transactions on Machine Learning Research · 摘要：Sparse PCA finds low-dimensional structure that loads on few features. Existing methods couple learning and feature selection by applying non-differentiable penalties that force retain-or-zero decisions during optimization. This eliminates features before the optimizer has established which ones matter, limits scalability to serial coordinate-update solvers, and fails when components share support. We introduce DROSS-PCA (Differentiable RObust Scalable Sparse PCA), which decouples learning and feature selection. It learns by optimizing a fully differentiable objective combining robust reconst… · 篇幅：Long submission (more than 12 pages of main content) · 代码：https://github.com/nihcompmed/DROSS-PCA · OpenReview ID：CZwpmO1irdManu Aggarwal, Vipul Periwal
  • 39
    SQS: Bayesian DNN Compression through Sparse Quantized Sub-distributions
    期刊：Transactions on Machine Learning Research · 摘要：Compressing large-scale neural networks is essential for deploying models on resource-constrained devices. Most existing methods adopt weight pruning or low-bit quantization individually, often resulting in suboptimal compression rates to preserve acceptable performance drops. We introduce a unified fraework for simultaneous pruning and low-bit quantization via Bayesian variational learning (SQS), which achieves higher compression rates than prior baselines while maintaining comparable performance. The key idea is to employ a spike-and-slab prior to induce sparsity and model quantized weights… · 篇幅：Regular submission (no more than 12 pages of main content) · 代码：https://github.com/comeusr/SQS_TMLR · OpenReview ID：3nZb43fvAQZiyi Wang, Nan Jiang, Guang Lin et al.
  • 40
    Characterizing Correlation for Time Series Modeling Strategies: Evidence from SOFTS Variants
    期刊：Transactions on Machine Learning Research · 摘要：Channel-dependent (CD) and channel-independent (CI) strategies are competing inductive biases in long-term time series forecasting. Empirical evidence indicates that CD strategies benefit from stronger inter-channel correlations. Yet, the nature of these correlations remains unexplored, and increasingly sophisticated CI, CD, and hybrid architectures continue to be proposed without a clear understanding of when each is preferable. This study takes the hybrid CI-CD Series-cOre Fused Time Series forecaster (SOFTS) and pushes it to the extremes. We introduce the fully CD Channel Mixer SOFTS (C-SO… · 篇幅：Long submission (more than 12 pages of main content) · 代码：https://github.com/stevenyirenkyi/SOFTS-Variants · OpenReview ID：AmJiOQu9qhSteven Yirenkyi, Cyril Dziedzorm Boateng, KWABENA OWUSU-AGYEMANG et al.
  • 41
    Support Vector Machines: A More Certain Estimate of Uncertainty in Regression Tasks
    期刊：Transactions on Machine Learning Research · 摘要：This paper explores the potential of Support Vector Machines (SVMs) from the lens of uncertainty quantification (UQ), for regression and forecasting tasks. Unlike Neural Network (NN) and Random Forest (RF) based approaches, which are commonly used for UQ estimation, SVM-based methods are more stable and sparse, and offer well-defined optimal solutions. This makes SVM models often exhibit much lower model uncertainty than alternative learning models. However, there is only limited literature addressing UQ in SVM-based prediction, particularly in the regression setting. We first provide a compr… · 篇幅：Long submission (more than 12 pages of main content) · OpenReview ID：ZdzIDoGeOCPritam Anand
  • 42
    Stacked Feynman-Kac: A Generalised Method for Within-Timestep Sampling of Intermediate Distributions for Diffusion Models
    期刊：Transactions on Machine Learning Research · 摘要：Generative diffusion models have proven highly effective for both unconditional generation and conditional tasks. Sequential Monte Carlo (SMC) methods provide a principled framework for conditional sampling from diffusion priors, with applications including inpainting, super-resolution, and de-blurring. However, existing SMC-based approaches typically only propagate particles across timesteps and does not sufficiently explore the intermediate target distributions. To address this, we introduce the Stacked Feynman-Kac (SFK) algorithm, which enables sampling of intermediate distribution estimat… · 篇幅：Long submission (more than 12 pages of main content) · 代码：https://github.com/Armad1ll0/Stacked_Feynman_Kac · OpenReview ID：AVVFwOIfLYAndrew Millard, Johan Alenlöv
  • 43
    Not All Objectives Are Born Equal: Priority-Constrained Descent for Hierarchical Multi-Objective Optimization
    期刊：Transactions on Machine Learning Research · 摘要：Deep learning problems rarely involve objectives that are equal in importance. A primary objective defines the goal, whilst secondary objectives, such as sparsity, compression, or robustness, constrain the solution. While existing multi-objective methods have proven effective in practice, they have a clear symmetry problem and neglect the inherent objective hierarchy built into these objective spaces. We introduce Priority-Constrained Descent (PCD), a gradient-based optimization framework designed to explicitly exploit hierarchical objective structures. PCD preserves the direction of primary… · 篇幅：Long submission (more than 12 pages of main content) · 代码：https://daravaram.github.io/PCD/ · OpenReview ID：HT01yGHLEtDara Varam, Mohamed I. AlHajri
  • 44
    A variational approach to dimension-free self-normalized concentration
    期刊：Transactions on Machine Learning Research · 摘要：We study the self-normalized concentration of vector-valued stochastic processes. We focus on bounds for ``sub-$\psi$'' processes, a well-known and quite general class of process that encompasses a wide variety of well-known tail conditions (including sub-exponential, sub-Gaussian, sub-gamma, sub-Poisson, and several heavy-tailed settings without a moment generating function such as symmetric or bounded 2nd or 3rd moments). Our results recover and generalize the influential bound of de la Peña et al. (2004) (proved again in Abbasi-Yadkori et al. 2011) in the sub-Gaussian case. Further, we fil… · 篇幅：Regular submission (no more than 12 pages of main content) · 代码：https://github.com/bchugg/sn-concentration · OpenReview ID：XX5vENsuwxBen Chugg, Aaditya Ramdas
  • 45
    Large Pretraining Datasets Don't Guarantee Robustness after Fine-Tuning in Image Classification
    期刊：Transactions on Machine Learning Research · 摘要：Large-scale pretrained models are widely leveraged as foundations for learning new specialized tasks via fine-tuning, with the goal of maintaining the general performance of the model while allowing it to gain new skills. A valuable goal for all such models is robustness: the ability to perform well on out-of-distribution (OOD) tasks. We assess whether fine-tuning preserves the overall robustness of the pretrained model in image classification, and observed that models pretrained on large datasets exhibited strong catastrophic forgetting and loss of OOD generalization. To systematically asses… · 篇幅：Regular submission (no more than 12 pages of main content) · OpenReview ID：VyVkIucjWUJaedong Hwang, Brian Cheung, Zhang-Wei Hong et al.
  • 46
    Regular Fourier Features for Nonstationary Gaussian Processes
    期刊：Transactions on Machine Learning Research · 摘要：Simulating a Gaussian process requires sampling from a high-dimensional Gaussian distribution, which scales cubically with the number of sample locations. Spectral methods address this challenge by exploiting the Fourier representation and treating the spectral density as a probability distribution suitable for Monte Carlo approximation. Although this probabilistic interpretation is valid for stationary processes, it is overly restrictive for the nonstationary case, where spectral densities are generally not probability measures. To avoid this limitation, we propose regular Fourier features f… · 篇幅：Long submission (more than 12 pages of main content) · 代码：https://github.com/anjawa/neural-spectral-gp · OpenReview ID：2eZhxVDAhRArsalan Jawaid, Abdullah Karatas, Jörg Seewig
  • 47
    MiLDEdit: Reasoning-Based Multi-Layer Design Document Editing
    期刊：Transactions on Machine Learning Research · 摘要：Real-world design documents (e.g., posters) are inherently multi-layered, combining decoration, text, and images. Editing them from natural-language instructions requires fine-grained, layer-aware reasoning to identify relevant layers and coordinate modifications. Prior work largely overlooks multi-layer design document editing, focusing instead on single-layer image editing or multi-layer generation, which assume a flat canvas and lack the reasoning needed to determine what and where to modify. To address this gap, we introduce the Multi-Layer Document Editing Agent (MiLDEAgent), a reasoning… · 篇幅：Regular submission (no more than 12 pages of main content) · OpenReview ID：G3l9DAz5uHZihao Lin, Wanrong Zhu, Jiuxiang Gu et al.
  • 48
    EditProp: Consistent Video Style Transfer by Editing Propagation
    期刊：Transactions on Machine Learning Research · 摘要：Video style transfer, which aims to transfer a source video into another video with a different appearance while preserving its original structure, plays an important role in the video production industry. Existing methods often edit the first frame with an image editing tool, and feed it into an image-to-video generation model with source video guidance to generate the edited video. Although such a paradigm enables users to perform creative video editing with powerful image editing tools, it relies heavily on the native propagation capability of the video generation model, which can be limit… · 篇幅：Regular submission (no more than 12 pages of main content) · OpenReview ID：WA0ApsjWQbGuangzhi Wang, Xiaoyu Li, Xiaodong Cun et al.
  • 49
    Greedy Decoding Is Not Precision-Invariant: Cross-Precision Output Divergence in LLM Inference
    期刊：Transactions on Machine Learning Research · 摘要：Greedy decoding from large language models is commonly treated as deterministic. We show it is not precision-invariant: the same model, prompt, and decoding algorithm produce different outputs in BF16 versus FP16 on identical hardware. Across our evaluations of six models (1.1B–7B parameters, four families; divergence additionally characterised at 12B) and three benchmarks, 49–100% of prompts diverge; a single token flip often cascades into trajectory-level divergence. We develop an empirical error-propagation analysis and find that 22 layers of accumulated body error do not distinguish flipp… · 篇幅：Long submission (more than 12 pages of main content) · OpenReview ID：QDOKyg7a5eGaoyuan Du, Anam Nawaz Khan, Rex Zhou et al.
  • 50
    Economies of Open Intelligence: Tracing Power & Participation in the Model Ecosystem
    期刊：Transactions on Machine Learning Research · 摘要：Since 2019, the Hugging Face Model Hub has been the primary global platform for sharing open weight AI models. By releasing a dataset of the complete history of weekly model downloads (June 2020–August 2025) alongside model metadata, we provide the most rigorous examination to-date of download-based concentration dynamics and evolving characteristics in the open model ecosystem. Our analysis spans 851,000 models, over 200 aggregated attributes per model, and 2.2B downloads, establishing persistent scientific infrastructure for measuring how AI capability, influence, and participation diffuse… · 篇幅：Regular submission (no more than 12 pages of main content) · OpenReview ID：lYvDFMUcwvShayne Longpre, Christopher Akiki, Campbell S. Lund et al.
共 50 条热搜查看完整榜单

Phys.org · 实时热榜

14分钟前更新
  • 01
    Why AI has trouble predicting the fury of hurricane intensity
    Artificial intelligence has revolutionized weather forecasting in just a few years, with global AI weather models now able to produce forecasts that rival some of the world's best physics-based prediction systems.
  • 02
    NEET with a disability: What's holding back visually impaired young people who want to work
    I have been fortunate in my research to follow 80 young people with vision impairment over 11 years. I tracked their journeys from school into adulthood, speaking with them at regular intervals to learn what had happened since we last spoke, what had gone well and what could have been improved.
  • 03
    Looking for #womenshealth information on Instagram? 4 out of 10 posts are ads
    A woman in her 40s scrolls through Instagram. Between photos of friends, recipes and vacation photos, she comes across a video about perimenopause. The woman on screen explains that perimenopause is to blame for feeling tired, having trouble remembering things, struggling to sleep or gaining weight.
  • 04
    Black hole jets reach far beyond galaxies' visible edges, potentially deciding their fate
    Galaxies are enormous and hold hundreds of billions of stars. These stars form from cold, dense gas. Every large galaxy, including the Milky Way, is wrapped in a huge envelope of gas called the circumgalactic medium, or CGM. This reservoir of raw material stretches 10–20 times the size of the visible portion of the galaxy.
  • 05
    Why earwigs infected by parasitic worms head towards water
    Earwigs, small, elongated insects with distinctive pincers at the rear of their abdomens, have little reason to enter open water. While many earwig species are attracted to damp environments, they typically avoid open water. Yet past studies have found that European earwigs infected with a parasitic roundworm called Mermis nigrescens tend to head toward open water. The molecular and biological reasons for this unusual behavior have not been fully uncovered.
  • 06
    Pumpkin-derived enzyme weakens peanut proteins' binding to allergy antibodies in lab tests
    Peanut allergy is one of the most serious food allergies and can lead to sudden, life-threatening reactions. Researchers from Wroclaw University of Environmental and Life Sciences, in collaboration with Wroclaw Medical University, investigated whether a natural enzyme obtained from figleaf gourd (Cucurbita ficifolia) could reduce recognition of peanut proteins by antibodies involved in allergic reactions.
  • 07
    Engineers teach spacecraft to 'dream' their way to the space station
    Docking with the ISS may seem simple. However, actually doing so shows how difficult orbital mechanics can be. It's like traveling down a highway at 28,000 km/hr (17,000 mph) and parallel parking in an open garage on a multibillion-dollar laboratory traveling at the same speed. If you try to accelerate forward, you actually drift up, and there's no air friction to naturally slow you down. Oh, and if you hit the lab, everyone aboard both your craft and the station dies, and the resulting debris f
  • 08
    Sparse layers of 10-nanometer silver particles boost CO production from CO₂
    Electrolysis can reduce CO2 to CO, a raw material for chemical products such as fuels. Within the GreenQUEST project, an international team led by HZB chemist Prashanth Menezes has systematically investigated catalyst layers made of silver nanoparticles, varying both particle size and density.
  • 09
    A real-time look inside spacecraft heat shields during extreme heat conditions
    When a spacecraft reenters Earth's atmosphere at hypersonic velocities, its protective heat shield faces extreme conditions, including temperatures beyond 3,000 degrees Fahrenheit (1,650 degrees Celsius). To survive, the shield relies on specialized materials that absorb heat as they degrade in a controlled process called ablation, sacrificing the shield's outer layers to protect the vehicle and crew within.
  • 10
    Crop protection: A new method for understanding insect sexual communication
    A team of INRAE scientists, in collaboration with Université Côte d'Azur and Nanjing Agricultural University in China, has used an innovative AI-based method to identify the sex pheromone of the lily moth1 and the olfactory receptors associated with it. The insect pest, whose caterpillars feed on these plants, is found in Asia and Oceania. This approach to identifying previously unknown molecules opens up new possibilities for identifying pheromones in other species. These molecules play a key r
  • 11
    Lead isotopes trace Malawi's mysterious copper ingot hoard to Congo's Kipushi mine
    In 1967, a Malawian farmer stumbled upon a hidden hoard of eight copper ingots buried beneath a tree. Now, more than half a century later, scientists have traced the metal to its source: a copper mine nearly 800 km (500 miles) away.
  • 12
    Molecular 'self-destruct' switch discovered: How stressed cells decide between survival and death
    A new study from Ben-Gurion University of the Negev (BGU) has identified a molecular switch that controls whether stressed human cells activate a survival mechanism or trigger programmed cell death (apoptosis). The findings, published in Redox Biology, provide insights that could help overcome chemotherapy resistance in aggressive tumors.
  • 13
    Digital textbooks significantly boost students' intrinsic learning motivation
    A research team from The Education University of Hong Kong (EdUHK) and East China Normal University (ECNU) has published a meta-analysis demonstrating the impact of digital textbooks (DTs) on student motivation. Synthesizing 27 experimental and quasi-experimental studies conducted between 2000 and 2024 with 3,590 participants, the study confirms that DTs significantly enhance students' intrinsic learning motivation compared with traditional paper textbooks.
  • 14
    Millions in the US Northeast brace for a powerful storm
    More than 40 million people in the northeastern United States braced Friday for heavy rain and strong winds as an unusually powerful storm for the season was forecast to crash into the region.
  • 15
    Canals full, roads submerged as Bangkok declares flood disaster
    Bangkok authorities declared a flood disaster across the Thai capital on Saturday after almost 48 hours of heavy rain left canals full, roads submerged and residents wading through deep water.
  • 16
    Indonesia fires threaten critically endangered orangutan: IUCN
    Wildfires that tore through Indonesian Borneo's vast tropical forests have increased pressure on critically endangered Bornean orangutans, the International Union for Conservation of Nature (IUCN) said Saturday.
  • 17
    Guatemala ensures survival of vulnerable 'fossil' fish
    In a freshwater canal surrounded by mangroves in southern Guatemala, hundreds of baby tropical gars—a torpedo-shaped fish with an alligator-like snout dubbed the "fossil fish"—are released into the wild.
  • 18
    A 'library' of bacteria-killing viruses: Harnessing the power of phages
    An estimated 38 trillion bacteria live in the human body—and only a small fraction are harmful to our health. However, the harmful ones can cause serious illness, especially when they develop resistance to antibiotics.
  • 19
    Climate crisis is affecting children's nutrition
    Wildfires, floods and other extreme weather events are no longer the only ways climate change affects children around the world. Increasingly, it also affects their diets. Droughts reduce agricultural production, floods cut off roads and disrupt supply chains, and rising prices limit families' access to fruits, vegetables and other fresh foods.
  • 20
    Breaking and entering: How parasites prepare host cells for invasion
    Before the parasite Toxoplasma gondii can invade a cell, it has to know it has reached the right place. New research from the Whitehead Institute reveals that the parasite doesn't simply recognize a ready-made doorway into the cell. Instead, it helps create the conditions for its own entry.
  • 21
    How cells rapidly detect viruses that fuse with their membranes
    Researchers have figured out a missing piece of the puzzle of how human cells detect invading viruses and mount an immune response within an hour.
  • 22
    Critically endangered species in the heart of Africa: Investigating Hooded Vultures in Chad
    Slaughterhouses are places most people never want to go, and might seem like unlikely places to search for rare or endangered species. In West Africa, however, they are among the last places one can find a critically endangered species, the Hooded Vulture, which historically thrived as a human commensal species, living in close proximity to people and contributing to waste disposal.
  • 23
    Mitochondrial protein complexes: A single amino acid can jeopardize their stability
    Removing a single amino acid can determine whether new proteins assemble into stable complexes within mitochondria. An enzyme carries out this precise trimming at the start of mitochondrial proteins. Without it, numerous protein complexes lose their stability, according to a study led by Dr. Nora Vögtle of Heidelberg University and Dr. Pitter Huesgen of the University of Freiburg. In their work published in Nature Structural & Molecular Biology, the scientists identified a previously unknown mec
  • 24
    Blood pressure enzyme reveals second job: Making sulfur rings for cellular antioxidants
    Antioxidants such as vitamins C and E protect cells by neutralizing free radicals, such as reactive oxygen species, before they can damage genetic material or cell membranes. Such damage can contribute to the development of neurodegenerative diseases. Antioxidants therefore help prevent cellular damage.
  • 25
    Natural proteins can guide active ingredients to diseased cells
    After taking a pill, the drugs it contains are distributed throughout the body, reaching the tissues and cells affected by disease and performing their intended function. However, they also enter healthy cells, where they often cause more harm than good.
  • 26
    How a century-old solvent can enable efficient amide synthesis
    Amide bonds are chemical linkages essential to biology, medicine and materials chemistry. Peptide bonds, a class of amide bonds, link amino acids to form proteins. Amide bonds are also important in medicines and polymers. Efficient synthesis of amide bonds remains a major focus in green chemistry.
  • 27
    Ultrathin materials could make quantum light circuits programmable
    Quantum photonics could be a pivotal part of future quantum technology if the right materials can be created, a new review paper has found.
  • 28
    An overlooked era of cosmology is stealing the show in the quest to understand the cosmos
    Could you imagine if pagers made a comeback? Sure, your smartphone would still be your notepad-flashlight-camera-personal assistant-MP3 player-wallet-social media browser. But imagine if pagers made it so easy to do that one important thing—pinging your friends and associates—that we all brought them back into our lives, or adopted them for the first time.
  • 29
    Laser light enables detection of colorectal cancer biomarker in blood within minutes
    Researchers often use antibodies to detect specific molecules or biological markers because they are remarkably precise at recognizing their targets. However, finding the faint signals of cancer in blood samples is a challenge even for the most reliable techniques.
  • 30
    New tools pinpoint where urban trees could provide the most relief from heat
    Cities trying to cool sweltering streets often set broad tree canopy targets. But those targets can overlook the reality that not all shade is equally useful, and not every neighborhood has room for more trees.
共 30 条热搜查看完整榜单

PubMed · 实时热榜

4小时前更新
共 50 条热搜查看完整榜单
  • 01
    Gravity Seems Holographic. What Does That Mean for Reality?
    The biggest breakthrough in modern theoretical physics is the discovery that gravity can collapse the dimensions of space. Physicists don’t yet understand the implications. The post Gravity Seems Holographic. What Does That Mean for Reality? first appeared on Quanta MagazineCharlie Wood
  • 02
    Biology Might Not Be Quantum, but Its Math Is Quantumlike
    Scientists have a history of trying — and failing — to link biology and quantum mechanics. The real connection between them may be in the math. The post Biology Might Not Be Quantum, but Its Math Is Quantumlike first appeared on Quanta MagazineElise Cutts
  • 03
    How Virus-like ‘Jumping Genes’ Became Our Partners in Evolution
    Half of our genome is made of transposons — snips of DNA that can move and copy themselves. But they’re more than parasites or genetic junk. The post How Virus-like ‘Jumping Genes’ Became Our Partners in Evolution first appeared on Quanta MagazineJake Buehler
  • 04
    Mathematicians Build Long-Awaited Graph Sandwich
    The proof of a decades-old conjecture has given researchers a new way to understand complex networks. The post Mathematicians Build Long-Awaited Graph Sandwich first appeared on Quanta MagazinePaulina Rowińska
  • 05
    Where Does the Quantum World End and Ours Begin?
    Jonathan Halliwell explains how quantum decoherence is key to understanding how we transition from a world with a wave-like nature of matter and energy to the classical macroscopic world that we’re used to. The post Where Does the Quantum World End and Ours Begin? first appeared on Quanta MagazineSteven Strogatz and Janna Levin
共 5 条热搜查看完整榜单
共 10 条热搜查看完整榜单
共 10 条热搜查看完整榜单
知识热榜 - 今日热门话题与学术资讯 - 第 2 页 | 什么火了