Nature Machine Intelligence最新文献_第5页

Designing metamaterials with programmable nonlinear responses and geometric constraints in graph space 在图空间中设计具有可编程非线性响应和几何约束的超材料

IF 23.9 1区计算机科学

Nature Machine Intelligence Pub Date : 2025-07-22 DOI: 10.1038/s42256-025-01067-x

Marco Maurizi, Derek Xu, Yu-Tong Wang, Desheng Yao, David Hahn, Mourad Oudich, Anish Satpati, Mathieu Bauchy, Wei Wang, Yizhou Sun, Yun Jing, Xiaoyu Rayne Zheng

{"title":"Designing metamaterials with programmable nonlinear responses and geometric constraints in graph space","authors":"Marco Maurizi, Derek Xu, Yu-Tong Wang, Desheng Yao, David Hahn, Mourad Oudich, Anish Satpati, Mathieu Bauchy, Wei Wang, Yizhou Sun, Yun Jing, Xiaoyu Rayne Zheng","doi":"10.1038/s42256-025-01067-x","DOIUrl":"10.1038/s42256-025-01067-x","url":null,"abstract":"Advances in data-driven design and additive manufacturing have substantially accelerated the development of truss metamaterials—three-dimensional truss networks—offering exceptional mechanical properties at a fraction of the weight of conventional solids. While existing design approaches can generate metamaterials with target linear properties, such as elasticity, they struggle to capture complex nonlinear behaviours and to incorporate geometric and manufacturing constraints—including defects—crucial for engineering applications. Here we present GraphMetaMat, an autoregressive graph-based framework capable of designing three-dimensional truss metamaterials with programmable nonlinear responses, originating from hard-to-capture physics such as buckling, frictional contact and wave propagation, along with arbitrary geometric constraints and defect tolerance. Integrating graph neural networks, physics biases, imitation learning, reinforcement learning and tree search, we show that GraphMetaMat can target stress–strain curves across four orders of magnitude and vibration transmission responses with varying attenuation gaps, unattainable by previous methods. We further demonstrate the use of GraphMetaMat for the inverse design of novel material topologies with tailorable high-energy absorption and vibration damping that outperform existing polymeric foams and phononic crystals, potentially suitable for protective equipment and electric vehicles. This work sets the stage for the automatic design of manufacturable, defect-tolerant materials with on-demand functionalities. Maurizi et al. introduce GraphMetaMat, a graph-based AI framework for designing 3D metamaterials with programmable nonlinear responses, enabling the inverse design of new structural and acoustic behaviours despite fabrication defects and limits.","PeriodicalId":48533,"journal":{"name":"Nature Machine Intelligence","volume":"7 7","pages":"1023-1036"},"PeriodicalIF":23.9,"publicationDate":"2025-07-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144677427","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Unifying multi-sample network inference from prior knowledge and omics data with CORNETO 用CORNETO统一先验知识和组学数据的多样本网络推理

IF 23.9 1区计算机科学

Nature Machine Intelligence Pub Date : 2025-07-22 DOI: 10.1038/s42256-025-01069-9

Pablo Rodriguez-Mier, Martin Garrido-Rodriguez, Attila Gabor, Julio Saez-Rodriguez

{"title":"Unifying multi-sample network inference from prior knowledge and omics data with CORNETO","authors":"Pablo Rodriguez-Mier, Martin Garrido-Rodriguez, Attila Gabor, Julio Saez-Rodriguez","doi":"10.1038/s42256-025-01069-9","DOIUrl":"10.1038/s42256-025-01069-9","url":null,"abstract":"Understanding biological systems requires methods that extract interpretable insights from omics data. Networks offer a natural abstraction by representing molecules as vertices and their interactions as edges, providing a foundation for constructing context-specific models tailored to particular conditions—an essential step in many biological analyses. Most existing approaches fall into one of two categories: machine learning methods, which offer strong predictive power but lack interpretability and require large datasets, and knowledge-based methods, which are more interpretable but designed for analysing individual samples and difficult to generalize. Here we present CORNETO, a unified mathematical framework that generalizes a wide variety of methods that learn biological networks from omics data and prior knowledge. CORNETO reformulates these methods as mixed-integer optimization problems using network flows and structured sparsity, enabling joint inference across multiple samples. This improves the discovery of both shared and sample-specific molecular mechanisms while yielding sparser, more interpretable solutions. CORNETO supports a range of prior knowledge structures, including undirected, directed and signed (hyper)graphs. It extends a broad class of approaches, ranging from Steiner trees to flux balance analysis, within a unified optimization-based interface. We demonstrate CORNETO’s utility across diverse biological contexts, including signalling, metabolism and integration with biologically informed deep learning. We provide CORNETO as an open-source Python library for flexible network modelling. CORNETO is a unified mathematical framework and software that integrates prior knowledge with omics data to jointly infer context-specific signalling, metabolic and protein networks across multiple samples, boosting interpretability and accuracy.","PeriodicalId":48533,"journal":{"name":"Nature Machine Intelligence","volume":"7 7","pages":"1168-1186"},"PeriodicalIF":23.9,"publicationDate":"2025-07-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.nature.comhttps://www.nature.com/articles/s42256-025-01069-9.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144677710","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Bioinspired trajectory modulation for effective slip control in robot manipulation 仿生轨迹调制在机器人操作中的有效滑移控制

IF 23.9 1区计算机科学

Nature Machine Intelligence Pub Date : 2025-07-22 DOI: 10.1038/s42256-025-01062-2

Kiyanoush Nazari, Willow Mandil, Marco Santello, Seongjun Park, Amir Ghalamzan-E

{"title":"Bioinspired trajectory modulation for effective slip control in robot manipulation","authors":"Kiyanoush Nazari, Willow Mandil, Marco Santello, Seongjun Park, Amir Ghalamzan-E","doi":"10.1038/s42256-025-01062-2","DOIUrl":"10.1038/s42256-025-01062-2","url":null,"abstract":"Ensuring a stable grasp during robotic manipulation is essential for dexterous and reliable performance. Traditionally, slip control has relied on grip force modulation. Here we show that trajectory modulation provides an effective alternative for slip prevention in certain robotic manipulation tasks. We develop and compare a slip control policy based on trajectory modulation with a conventional grip-force-based approach. Our results demonstrate that trajectory modulation can significantly outperform grip force control in specific scenarios, highlighting its potential as a robust slip control strategy. Furthermore, we show that, similar to humans, incorporating a data-driven action-conditioned forward model within a model predictive control framework is key to optimizing trajectory modulation for slip prevention. These findings introduce a predictive control framework leveraging trajectory adaptation, offering a new perspective on slip mitigation. This approach enhances grasp stability in dynamic and unstructured environments, improving the adaptability of robotic systems across various applications. When a robot grips and moves a delicate object, it can slip from grasp. Instead of gripping the object with more force, a method is proposed here to move the object in a way that prevents slippage.","PeriodicalId":48533,"journal":{"name":"Nature Machine Intelligence","volume":"7 7","pages":"1119-1128"},"PeriodicalIF":23.9,"publicationDate":"2025-07-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.nature.comhttps://www.nature.com/articles/s42256-025-01062-2.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144677428","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Combining grasping and rotation with a spherical robot hand mechanism 结合抓握和旋转的球形机械手机构

IF 23.9 1区计算机科学

Nature Machine Intelligence Pub Date : 2025-07-21 DOI: 10.1038/s42256-025-01039-1

Vatsal V. Patel, Aaron M. Dollar

{"title":"Combining grasping and rotation with a spherical robot hand mechanism","authors":"Vatsal V. Patel, Aaron M. Dollar","doi":"10.1038/s42256-025-01039-1","DOIUrl":"10.1038/s42256-025-01039-1","url":null,"abstract":"Object reorientation is a key functionality in dexterous manipulation tasks, such as turning a doorknob. This is usually done on robot arms with a simple gripper and a three-degrees-of-freedom wrist. However, wrists are mechanically complex, and the wrist axes are often far away from the grasped object, resulting in coupled translations that need to be compensated with awkward whole-arm motions. We present a robot hand mechanism based on a spherical parallel architecture that can both grasp and rotate a wide range of objects in all three axes, combining much of the function of traditional wrists and grippers. The hand mechanism allows for pure spherical rotations of the grasped object about a known fixed point close to the object, thereby avoiding parasitic translations and inefficient arm motions. This point also stays fixed with respect to the hand, and is independent of the object shape, pose or initial grasp. We detail the spherical parallel design and workspace model of the wrist-like Sphinx hand, validate its performance for lower-degrees-of-freedom robot arms without traditional wrists and show that it can accurately rotate the grasped objects over large angles with basic open-loop control. Developing robot hands for unstructured human environments is a major challenge. A robotic hand that combines grasping and wrist-like rotation in one mechanism for more efficient and versatile object manipulation is presented.","PeriodicalId":48533,"journal":{"name":"Nature Machine Intelligence","volume":"7 7","pages":"999-1009"},"PeriodicalIF":23.9,"publicationDate":"2025-07-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144670044","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Advancing biomolecular understanding and design following human instructions 推进生物分子的理解和设计遵循人类的指示

IF 23.9 1区计算机科学

Nature Machine Intelligence Pub Date : 2025-07-21 DOI: 10.1038/s42256-025-01064-0

Xiang Zhuang, Keyan Ding, Tianwen Lyu, Yinuo Jiang, Xiaotong Li, Zhuoyi Xiang, Zeyuan Wang, Ming Qin, Kehua Feng, Jike Wang, Qiang Zhang, Huajun Chen

{"title":"Advancing biomolecular understanding and design following human instructions","authors":"Xiang Zhuang, Keyan Ding, Tianwen Lyu, Yinuo Jiang, Xiaotong Li, Zhuoyi Xiang, Zeyuan Wang, Ming Qin, Kehua Feng, Jike Wang, Qiang Zhang, Huajun Chen","doi":"10.1038/s42256-025-01064-0","DOIUrl":"10.1038/s42256-025-01064-0","url":null,"abstract":"Understanding and designing biomolecules, such as proteins and small molecules, is central to advancing drug discovery, synthetic biology and enzyme engineering. Recent breakthroughs in artificial intelligence have revolutionized biomolecular research, achieving remarkable accuracy in biomolecular prediction and design. However, a critical gap remains between artificial intelligence’s computational capabilities and researchers’ intuitive goals, particularly in using natural language to bridge complex tasks with human intentions. Large language models have shown potential to interpret human intentions, yet their application to biomolecular research remains nascent due to challenges including specialized knowledge requirements, multimodal data integration, and semantic alignment between natural language and biomolecules. To address these limitations, we present InstructBioMol, a large language model designed to bridge natural language and biomolecules through a comprehensive any-to-any alignment of natural language, molecules and proteins. This model can integrate multimodal biomolecules as the input, and enable researchers to articulate design goals in natural language, providing biomolecular outputs that meet precise biological needs. Experimental results demonstrate that InstructBioMol can understand and design biomolecules following human instructions. In particular, it can generate drug molecules with a 10% improvement in binding affinity and design enzymes that achieve an enzyme–substrate pair prediction score of 70.4. This highlights its potential to transform real-world biomolecular research. InstructBioMol, a multimodal large language model that achieves any-to-any alignment between human instructions and biomolecules, can effectively leverage natural language to connect complex biomolecular tasks with human intentions.","PeriodicalId":48533,"journal":{"name":"Nature Machine Intelligence","volume":"7 7","pages":"1154-1167"},"PeriodicalIF":23.9,"publicationDate":"2025-07-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144669780","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Mapping T helper cell targets with deep learning 用深度学习映射T辅助细胞目标

IF 23.9 1区计算机科学

Nature Machine Intelligence Pub Date : 2025-07-17 DOI: 10.1038/s42256-025-01081-z

Yuan Liu, Leng Han

引用次数: 0

Machine learning modelling for multi-order human visual motion processing 多阶人类视觉运动处理的机器学习建模

IF 23.9 1区计算机科学

Nature Machine Intelligence Pub Date : 2025-07-15 DOI: 10.1038/s42256-025-01068-w

Zitang Sun, Yen-Ju Chen, Yung-Hao Yang, Yuan Li, Shin’ya Nishida

{"title":"Machine learning modelling for multi-order human visual motion processing","authors":"Zitang Sun, Yen-Ju Chen, Yung-Hao Yang, Yuan Li, Shin’ya Nishida","doi":"10.1038/s42256-025-01068-w","DOIUrl":"10.1038/s42256-025-01068-w","url":null,"abstract":"Visual motion perception is a key function for agents interacting with their environment. Although recent advances in optical flow estimation using deep neural networks have surpassed human-level accuracy, a notable disparity remains. In addition to limitations in luminance-based first-order motion perception, humans can perceive motions in higher-order features—an ability lacking in conventional optical flow models that rely on intensity conservation law. To address this, we propose a dual-pathway model that mimics the cortical V1-MT motion processing pathway. It uses a trainable motion energy sensor bank and a recurrent graph network to process luminance-based motion and incorporates an additional sensing pathway with nonlinear preprocessing using a multilayer 3D CNN block to capture higher-order motion signals. We hypothesize that higher-order mechanisms are critical for estimating robust object motion in natural environments that contain complex optical fluctuations, for example, highlights on glossy surfaces. By training on motion datasets with varying material properties of moving objects, our dual-pathway model naturally developed the capacity to perceive multi-order motion as humans do. The resulting model effectively aligns with biological systems while generalizing both luminance-based and higher-order motion phenomena in natural scenes. Sun and colleagues present a brain-inspired dual-pathway model that learns to perceive both first- (luminance-based) and second-order (feature-based) motion, achieving human-like performance through training on naturalistic environments and diverse materials.","PeriodicalId":48533,"journal":{"name":"Nature Machine Intelligence","volume":"7 7","pages":"1037-1052"},"PeriodicalIF":23.9,"publicationDate":"2025-07-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.nature.comhttps://www.nature.com/articles/s42256-025-01068-w.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144630034","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Deep-learning-aided dismantling of interdependent networks 深度学习辅助拆解相互依赖的网络

IF 23.9 1区计算机科学

Nature Machine Intelligence Pub Date : 2025-07-14 DOI: 10.1038/s42256-025-01070-2

Weiwei Gu, Chen Yang, Lei Li, Jinqiang Hou, Filippo Radicchi

{"title":"Deep-learning-aided dismantling of interdependent networks","authors":"Weiwei Gu, Chen Yang, Lei Li, Jinqiang Hou, Filippo Radicchi","doi":"10.1038/s42256-025-01070-2","DOIUrl":"10.1038/s42256-025-01070-2","url":null,"abstract":"Identifying the minimal set of nodes whose removal breaks a complex network apart, also referred as the network dismantling problem, is a highly non-trivial task with applications in multiple domains. Whereas network dismantling has been extensively studied over the past decade, research has primarily focused on the optimization problem for single-layer networks, neglecting that many, if not all, real networks display multiple layers of interdependent interactions. In such networks, the optimization problem is fundamentally different as the effect of removing nodes propagates within and across layers in a way that can not be predicted using a single-layer perspective. Here we propose a dismantling algorithm named MultiDismantler, which leverages multiplex network representation and deep reinforcement learning to optimally dismantle multilayer interdependent networks. MultiDismantler is trained on small synthetic graphs; when applied to large, either real or synthetic, networks, it displays exceptional dismantling performance, clearly outperforming all existing benchmark algorithms. We show that MultiDismantler is effective in guiding strategies for the containment of diseases in social networks characterized by multiple layers of social interactions. Also, we show that MultiDismantler is useful in the design of protocols aimed at delaying the onset of cascading failures in interdependent critical infrastructures. The dismantling problem of removing the smallest set of nodes so that a given network breaks into disconnected components is hard to solve exactly. Gu and colleagues use deep reinforcement learning and a multiplex network representation to avoid the heavy computational cost.","PeriodicalId":48533,"journal":{"name":"Nature Machine Intelligence","volume":"7 8","pages":"1266-1277"},"PeriodicalIF":23.9,"publicationDate":"2025-07-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144622235","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A new eye on inherited retinal disease 对遗传性视网膜疾病的新认识

IF 23.9 1区计算机科学

Nature Machine Intelligence Pub Date : 2025-07-14 DOI: 10.1038/s42256-025-01079-7

Beau J. Fenner, Ching-Yu Cheng

引用次数: 0

Learning to adapt through bio-inspired gait strategies for versatile quadruped locomotion 学习通过仿生步态策略适应多用途四足运动

IF 23.9 1区计算机科学

Nature Machine Intelligence Pub Date : 2025-07-11 DOI: 10.1038/s42256-025-01065-z

Joseph Humphreys, Chengxu Zhou

{"title":"Learning to adapt through bio-inspired gait strategies for versatile quadruped locomotion","authors":"Joseph Humphreys, Chengxu Zhou","doi":"10.1038/s42256-025-01065-z","DOIUrl":"10.1038/s42256-025-01065-z","url":null,"abstract":"Legged robots must adapt their gait to navigate unpredictable environments, a challenge that animals master with ease. However, most deep reinforcement learning (DRL) approaches to quadruped locomotion rely on a fixed gait, limiting adaptability to changes in terrain and dynamic state. Here we show that integrating three core principles of animal locomotion-gait transition strategies, gait memory and real-time motion adjustments enables a DRL control framework to fluidly switch among multiple gaits and recover from instability, all without external sensing. Our framework is guided by biomechanics-inspired metrics that capture efficiency, stability and system limits, which are unified to inform optimal gait selection. The resulting framework achieves blind zero-shot deployment across diverse, real-world terrains and substantially outperforms baseline controllers. By embedding biological principles into data-driven control, this work marks a step towards robust, efficient and versatile robotic locomotion, highlighting how animal motor intelligence can shape the next generation of adaptive machines. Humphreys and Zhou present a learning-based robot control framework inspired by animal gait mechanisms that enables quadruped robots to generalize to diverse real-world terrains, transition between gaits and recover from instability.","PeriodicalId":48533,"journal":{"name":"Nature Machine Intelligence","volume":"7 7","pages":"1141-1153"},"PeriodicalIF":23.9,"publicationDate":"2025-07-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.nature.comhttps://www.nature.com/articles/s42256-025-01065-z.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144603471","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0