Pub Date : 2026-01-27DOI: 10.1038/s42256-026-01183-2
Agentic artificial intelligence (AI) frameworks are in vogue. However, implementing such systems in scientific research workflows requires clear motivations and explanations, given the risk of wasting computational as well as human resources.
{"title":"Multi-agent AI systems need transparency","authors":"","doi":"10.1038/s42256-026-01183-2","DOIUrl":"10.1038/s42256-026-01183-2","url":null,"abstract":"Agentic artificial intelligence (AI) frameworks are in vogue. However, implementing such systems in scientific research workflows requires clear motivations and explanations, given the risk of wasting computational as well as human resources.","PeriodicalId":48533,"journal":{"name":"Nature Machine Intelligence","volume":"8 1","pages":"1-1"},"PeriodicalIF":23.9,"publicationDate":"2026-01-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.nature.comhttps://www.nature.com/articles/s42256-026-01183-2.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146071442","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}
Pub Date : 2026-01-26DOI: 10.1038/s42256-025-01165-w
Yue Guo, Hao Zhang, Haitao Hu, Jialu Wu, Ji Cao, Chang-Yu Hsieh, Bo Yang
Systematic mapping of chemical perturbation responses is revolutionizing polypharmacological drug discovery, yet remains constrained by experimental scalability. Here we introduce XPert, a biologically informed dual-branch transformer model designed to model gene-specific perturbation effects and dose–time dynamics. The dual-branch architecture separately encodes pre-perturbation and post-perturbation cellular states, allowing the model to disentangle intrinsic transcriptional patterns from regulatory shifts triggered by perturbations. By leveraging context-aware gene network modelling, XPert overcomes the over-denoising issues inherent in dominant variational-autoencoder-based approaches, achieving 36.7% higher Pearson’s correlation coefficient and 78.2% lower mean square error in cold-cell generalization under single-dose–single-time scenarios. Through extension to multidose–multitime prediction, XPert precisely resolves pharmacodynamic trajectories and uncovers key molecular events underlying the drug effects. To address real-world data scarcity, we apply knowledge transfer from large-scale preclinical screens to clinical contexts, achieving up to 15.04% improvement in patient-specific response predictions. Furthermore, XPert provides mechanistic interpretability, as evidenced by the identification of clinically validated resistance biomarkers. A dual-branch framework that disentangles cell states from drug-induced regulatory shifts to predict transcriptional responses is presented. It captures nonlinear dose–time dynamics and excels in generalizing to unseen cellular contexts.
{"title":"Modelling drug-induced cellular perturbation responses with a biologically informed dual-branch transformer","authors":"Yue Guo, Hao Zhang, Haitao Hu, Jialu Wu, Ji Cao, Chang-Yu Hsieh, Bo Yang","doi":"10.1038/s42256-025-01165-w","DOIUrl":"10.1038/s42256-025-01165-w","url":null,"abstract":"Systematic mapping of chemical perturbation responses is revolutionizing polypharmacological drug discovery, yet remains constrained by experimental scalability. Here we introduce XPert, a biologically informed dual-branch transformer model designed to model gene-specific perturbation effects and dose–time dynamics. The dual-branch architecture separately encodes pre-perturbation and post-perturbation cellular states, allowing the model to disentangle intrinsic transcriptional patterns from regulatory shifts triggered by perturbations. By leveraging context-aware gene network modelling, XPert overcomes the over-denoising issues inherent in dominant variational-autoencoder-based approaches, achieving 36.7% higher Pearson’s correlation coefficient and 78.2% lower mean square error in cold-cell generalization under single-dose–single-time scenarios. Through extension to multidose–multitime prediction, XPert precisely resolves pharmacodynamic trajectories and uncovers key molecular events underlying the drug effects. To address real-world data scarcity, we apply knowledge transfer from large-scale preclinical screens to clinical contexts, achieving up to 15.04% improvement in patient-specific response predictions. Furthermore, XPert provides mechanistic interpretability, as evidenced by the identification of clinically validated resistance biomarkers. A dual-branch framework that disentangles cell states from drug-induced regulatory shifts to predict transcriptional responses is presented. It captures nonlinear dose–time dynamics and excels in generalizing to unseen cellular contexts.","PeriodicalId":48533,"journal":{"name":"Nature Machine Intelligence","volume":"8 1","pages":"96-112"},"PeriodicalIF":23.9,"publicationDate":"2026-01-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.nature.comhttps://www.nature.com/articles/s42256-025-01165-w.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146048232","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}
Mathematics olympiads are prestigious competitions in which both proposing and solving problems are highly honoured. Building artificial intelligence systems capable of addressing these olympiad-level challenges remains an open frontier in automated reasoning, particularly in geometry due to its unique blend of numerical precision and spatial intuition. Here we show that TongGeometry, a neuro-symbolic system using guided tree search, both discovers and proves olympiad-level geometry theorems. Within the same computational budget as existing state-of-the-art systems, TongGeometry establishes a larger repository of geometry theorems: 6.7 billion requiring auxiliary constructions, including 4.1 billion exhibiting geometric symmetry. Among these, three of TongGeometry’s discoveries were selected for regional mathematical olympiads, appearing in a national team qualifying exam in China and a top civil olympiad in the USA. Guided by fine-tuned large language models, TongGeometry solved all International Mathematical Olympiad geometry problems in the IMO-AG-30 benchmark, outperforming average top human competitors on this specific dataset. It also surpasses the existing state of the art across a broader spectrum of olympiad-level problems and requires only consumer-grade computing resources. These results demonstrate that TongGeometry operates as both a mathematical discoverer and a solver, becoming an artificial intelligence system to achieve this dual capability. The deployment of a preliminary system based on TongGeometry demonstrates practical applications and opens fresh possibilities for artificial-intelligence-assisted mathematical research and education. TongGeometry both solves and proposes olympiad-level geometry problems. It uses guided tree search to find hard but concise problems, making advanced mathematical reasoning more accessible.
{"title":"Proposing and solving olympiad geometry with guided tree search","authors":"Chi Zhang, Jiajun Song, Siyu Li, Yitao Liang, Yuxi Ma, Wei Wang, Yixin Zhu, Song-Chun Zhu","doi":"10.1038/s42256-025-01164-x","DOIUrl":"10.1038/s42256-025-01164-x","url":null,"abstract":"Mathematics olympiads are prestigious competitions in which both proposing and solving problems are highly honoured. Building artificial intelligence systems capable of addressing these olympiad-level challenges remains an open frontier in automated reasoning, particularly in geometry due to its unique blend of numerical precision and spatial intuition. Here we show that TongGeometry, a neuro-symbolic system using guided tree search, both discovers and proves olympiad-level geometry theorems. Within the same computational budget as existing state-of-the-art systems, TongGeometry establishes a larger repository of geometry theorems: 6.7 billion requiring auxiliary constructions, including 4.1 billion exhibiting geometric symmetry. Among these, three of TongGeometry’s discoveries were selected for regional mathematical olympiads, appearing in a national team qualifying exam in China and a top civil olympiad in the USA. Guided by fine-tuned large language models, TongGeometry solved all International Mathematical Olympiad geometry problems in the IMO-AG-30 benchmark, outperforming average top human competitors on this specific dataset. It also surpasses the existing state of the art across a broader spectrum of olympiad-level problems and requires only consumer-grade computing resources. These results demonstrate that TongGeometry operates as both a mathematical discoverer and a solver, becoming an artificial intelligence system to achieve this dual capability. The deployment of a preliminary system based on TongGeometry demonstrates practical applications and opens fresh possibilities for artificial-intelligence-assisted mathematical research and education. TongGeometry both solves and proposes olympiad-level geometry problems. It uses guided tree search to find hard but concise problems, making advanced mathematical reasoning more accessible.","PeriodicalId":48533,"journal":{"name":"Nature Machine Intelligence","volume":"8 1","pages":"84-95"},"PeriodicalIF":23.9,"publicationDate":"2026-01-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.nature.comhttps://www.nature.com/articles/s42256-025-01164-x.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146048367","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}
Pub Date : 2026-01-21DOI: 10.1038/s42256-025-01172-x
Sully F. Chen
Capturing the complexity of cardiovascular dynamics demands multiple monitoring modalities, each with inherent trade-offs. Diffusion-based modeling offers a promising route for synthesizing and generating cross-modal data.
{"title":"Jointly modeling cardiovascular biomarkers","authors":"Sully F. Chen","doi":"10.1038/s42256-025-01172-x","DOIUrl":"10.1038/s42256-025-01172-x","url":null,"abstract":"Capturing the complexity of cardiovascular dynamics demands multiple monitoring modalities, each with inherent trade-offs. Diffusion-based modeling offers a promising route for synthesizing and generating cross-modal data.","PeriodicalId":48533,"journal":{"name":"Nature Machine Intelligence","volume":"8 1","pages":"4-5"},"PeriodicalIF":23.9,"publicationDate":"2026-01-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146033302","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}
Artificial intelligence is revolutionizing scientific research, yet its growing integration into laboratory environments presents critical safety challenges. Large language models and vision language models now assist in experiment design and procedural guidance, yet their ‘illusion of understanding’ may lead researchers to overtrust unsafe outputs. Here we show that current models remain far from meeting the reliability needed for safe laboratory operation. We introduce LabSafety Bench, a comprehensive benchmark that evaluates models on hazard identification, risk assessment and consequence prediction across 765 multiple-choice questions and 404 realistic laboratory scenarios, encompassing 3,128 open-ended tasks. Evaluations on 19 advanced large language models and vision language models show that no model evaluated on hazard identification surpasses 70% accuracy. While proprietary models perform well on structured assessments, they do not show a clear advantage in open-ended reasoning. These results underscore the urgent need for specialized safety evaluation frameworks before deploying artificial intelligence systems in real laboratory settings. Large language models are starting to be used in safety-critical tasks such as controlling robots. Zhou et al. present LabSafety Bench, a benchmark evaluating the ability of large language models to identify hazards and assess laboratory risks.
{"title":"Benchmarking large language models on safety risks in scientific laboratories","authors":"Yujun Zhou, Jingdong Yang, Yue Huang, Kehan Guo, Zoe Emory, Bikram Ghosh, Amita Bedar, Sujay Shekar, Zhenwen Liang, Pin-Yu Chen, Tian Gao, Werner Geyer, Nuno Moniz, Nitesh V. Chawla, Xiangliang Zhang","doi":"10.1038/s42256-025-01152-1","DOIUrl":"10.1038/s42256-025-01152-1","url":null,"abstract":"Artificial intelligence is revolutionizing scientific research, yet its growing integration into laboratory environments presents critical safety challenges. Large language models and vision language models now assist in experiment design and procedural guidance, yet their ‘illusion of understanding’ may lead researchers to overtrust unsafe outputs. Here we show that current models remain far from meeting the reliability needed for safe laboratory operation. We introduce LabSafety Bench, a comprehensive benchmark that evaluates models on hazard identification, risk assessment and consequence prediction across 765 multiple-choice questions and 404 realistic laboratory scenarios, encompassing 3,128 open-ended tasks. Evaluations on 19 advanced large language models and vision language models show that no model evaluated on hazard identification surpasses 70% accuracy. While proprietary models perform well on structured assessments, they do not show a clear advantage in open-ended reasoning. These results underscore the urgent need for specialized safety evaluation frameworks before deploying artificial intelligence systems in real laboratory settings. Large language models are starting to be used in safety-critical tasks such as controlling robots. Zhou et al. present LabSafety Bench, a benchmark evaluating the ability of large language models to identify hazards and assess laboratory risks.","PeriodicalId":48533,"journal":{"name":"Nature Machine Intelligence","volume":"8 1","pages":"20-31"},"PeriodicalIF":23.9,"publicationDate":"2026-01-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145968784","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}
Pub Date : 2025-12-31DOI: 10.1038/s42256-025-01162-z
Erji Li, Yusong Wang, Lei Jin, Zheng Zong, Enze Zhu, Bao Wang, Qian Wang, Zongyin Yang, Wen-Yan Yin, Zhun Wei
In AI-driven metamaterials discovery, designing metasurfaces requires extrapolation to unexplored performance regimes to discover new structures. Here we introduce MetaAI, a physics-aware current-diffusion framework that synergizes spatial topologies and frequency-domain responses to discover non-intuitive metasurface architectures. Unlike conventional inverse design constrained by predefined specifications, MetaAI operates as a performance synthesizer by generating electrical current distributions that bridge electromagnetic performance and metasurface structures. This enables both in-distribution and out-of-distribution targets with diverse topologies. The core innovation of the proposed framework lies in its dual-domain diffusion module, which directly correlates meta-atom current mechanisms with electromagnetic behaviours to enable the discovery of structures with 17.2% wider operational bandwidths. We validate MetaAI across single-layer, multilayer and dynamically tunable metasurfaces, demonstrating out-of-distribution generalization across full-wave simulations and experimental prototypes. Metasurface design driven by AI faces challenges, such as extrapolation to unexplored performance regimes. MetaAI, a physics-aware current-diffusion framework, is introduced to advance metamaterial discovery from interpolation to extrapolation.
{"title":"Current-diffusion model for metasurface structure discoveries with spatial-frequency dynamics","authors":"Erji Li, Yusong Wang, Lei Jin, Zheng Zong, Enze Zhu, Bao Wang, Qian Wang, Zongyin Yang, Wen-Yan Yin, Zhun Wei","doi":"10.1038/s42256-025-01162-z","DOIUrl":"10.1038/s42256-025-01162-z","url":null,"abstract":"In AI-driven metamaterials discovery, designing metasurfaces requires extrapolation to unexplored performance regimes to discover new structures. Here we introduce MetaAI, a physics-aware current-diffusion framework that synergizes spatial topologies and frequency-domain responses to discover non-intuitive metasurface architectures. Unlike conventional inverse design constrained by predefined specifications, MetaAI operates as a performance synthesizer by generating electrical current distributions that bridge electromagnetic performance and metasurface structures. This enables both in-distribution and out-of-distribution targets with diverse topologies. The core innovation of the proposed framework lies in its dual-domain diffusion module, which directly correlates meta-atom current mechanisms with electromagnetic behaviours to enable the discovery of structures with 17.2% wider operational bandwidths. We validate MetaAI across single-layer, multilayer and dynamically tunable metasurfaces, demonstrating out-of-distribution generalization across full-wave simulations and experimental prototypes. Metasurface design driven by AI faces challenges, such as extrapolation to unexplored performance regimes. MetaAI, a physics-aware current-diffusion framework, is introduced to advance metamaterial discovery from interpolation to extrapolation.","PeriodicalId":48533,"journal":{"name":"Nature Machine Intelligence","volume":"8 1","pages":"59-69"},"PeriodicalIF":23.9,"publicationDate":"2025-12-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145894630","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}
Pub Date : 2025-12-31DOI: 10.1038/s42256-025-01161-0
Xiao Xiao, Le Zhang, Hongyu Zhao, Zuoheng Wang
Cell–cell interactions (CCI), driven by distance-dependent signalling, are important for tissue development and organ function. While imaging-based spatial transcriptomics offers unprecedented opportunities to unravel CCI at single-cell resolution, current analyses face challenges such as limited ligand–receptor pairs measured, insufficient spatial encoding and low interpretability. We present GITIII (graph inductive bias transformer for intercellular interaction investigation), a lightweight, interpretable, self-supervised graph transformer-based model that conceptualizes cells as words and their surrounding cellular neighbourhood as context that shapes the meaning or state of the central cell. GITIII infers CCI by examining the correlation between a cell’s state and its niche, enabling us to understand how sender cells influence the gene expression of receiver cells, visualize spatial CCI patterns, perform CCI-informed cell clustering and construct CCI networks. Applied to four spatial transcriptomics datasets across multiple species, organs and platforms, GITIII effectively identified and statistically interpreted CCI patterns in the brain and tumour microenvironments. Xiao et al. present GITIII, a lightweight and interpretable graph transformer for inferring spatial single-cell-level interactions and quantifying the influence of neighbouring cells on the gene expression of receiver cells in spatial transcriptomics.
{"title":"Inferring spatial single-cell-level interactions through interpreting cell state and niche correlations learned by self-supervised graph transformer","authors":"Xiao Xiao, Le Zhang, Hongyu Zhao, Zuoheng Wang","doi":"10.1038/s42256-025-01161-0","DOIUrl":"10.1038/s42256-025-01161-0","url":null,"abstract":"Cell–cell interactions (CCI), driven by distance-dependent signalling, are important for tissue development and organ function. While imaging-based spatial transcriptomics offers unprecedented opportunities to unravel CCI at single-cell resolution, current analyses face challenges such as limited ligand–receptor pairs measured, insufficient spatial encoding and low interpretability. We present GITIII (graph inductive bias transformer for intercellular interaction investigation), a lightweight, interpretable, self-supervised graph transformer-based model that conceptualizes cells as words and their surrounding cellular neighbourhood as context that shapes the meaning or state of the central cell. GITIII infers CCI by examining the correlation between a cell’s state and its niche, enabling us to understand how sender cells influence the gene expression of receiver cells, visualize spatial CCI patterns, perform CCI-informed cell clustering and construct CCI networks. Applied to four spatial transcriptomics datasets across multiple species, organs and platforms, GITIII effectively identified and statistically interpreted CCI patterns in the brain and tumour microenvironments. Xiao et al. present GITIII, a lightweight and interpretable graph transformer for inferring spatial single-cell-level interactions and quantifying the influence of neighbouring cells on the gene expression of receiver cells in spatial transcriptomics.","PeriodicalId":48533,"journal":{"name":"Nature Machine Intelligence","volume":"8 1","pages":"42-58"},"PeriodicalIF":23.9,"publicationDate":"2025-12-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145894631","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}
Pub Date : 2025-12-31DOI: 10.1038/s42256-025-01160-1
Alex Morehead, Nabin Giri, Jian Liu, Pawan Neupane, Jianlin Cheng
The effects of ligand binding on protein structures and their in vivo functions carry numerous implications for modern biomedical research and biotechnology development efforts such as drug discovery. Although several deep learning (DL) methods and benchmarks designed for protein–ligand docking have recently been introduced, so far no previous works have systematically studied the behaviour of the latest docking and structure prediction methods within the broadly applicable context of: (1) using predicted (apo) protein structures for docking (for example, for applicability to new proteins); (2) binding multiple (cofactor) ligands concurrently to a given target protein (for example, for enzyme design); and (3) having no previous knowledge of binding pockets (for example, for generalization to unknown pockets). To enable a deeper understanding of the real-world utility of docking methods, we introduce PoseBench, a comprehensive benchmark for broadly applicable protein–ligand docking. PoseBench enables researchers to rigorously and systematically evaluate DL methods for apo-to-holo protein–ligand docking and protein–ligand structure prediction using both primary ligand and multiligand benchmark datasets, the latter of which we introduce to the DL community. Empirically, using PoseBench, we find that: (1) DL cofolding methods generally outperform comparable conventional and DL docking baseline algorithms, but popular methods such as AlphaFold 3 are still challenged by prediction targets with new protein–ligand binding poses; (2) certain DL cofolding methods are highly sensitive to their input multiple sequence alignments, whereas others are not; and (3) DL methods struggle to strike a balance between structural accuracy and chemical specificity when predicting new or multiligand protein targets. Morehead et al. introduce the benchmark PoseBench and evaluate the strengths and limitations of current AI-based protein–ligand docking and structure prediction methods.
{"title":"Assessing the potential of deep learning for protein–ligand docking","authors":"Alex Morehead, Nabin Giri, Jian Liu, Pawan Neupane, Jianlin Cheng","doi":"10.1038/s42256-025-01160-1","DOIUrl":"10.1038/s42256-025-01160-1","url":null,"abstract":"The effects of ligand binding on protein structures and their in vivo functions carry numerous implications for modern biomedical research and biotechnology development efforts such as drug discovery. Although several deep learning (DL) methods and benchmarks designed for protein–ligand docking have recently been introduced, so far no previous works have systematically studied the behaviour of the latest docking and structure prediction methods within the broadly applicable context of: (1) using predicted (apo) protein structures for docking (for example, for applicability to new proteins); (2) binding multiple (cofactor) ligands concurrently to a given target protein (for example, for enzyme design); and (3) having no previous knowledge of binding pockets (for example, for generalization to unknown pockets). To enable a deeper understanding of the real-world utility of docking methods, we introduce PoseBench, a comprehensive benchmark for broadly applicable protein–ligand docking. PoseBench enables researchers to rigorously and systematically evaluate DL methods for apo-to-holo protein–ligand docking and protein–ligand structure prediction using both primary ligand and multiligand benchmark datasets, the latter of which we introduce to the DL community. Empirically, using PoseBench, we find that: (1) DL cofolding methods generally outperform comparable conventional and DL docking baseline algorithms, but popular methods such as AlphaFold 3 are still challenged by prediction targets with new protein–ligand binding poses; (2) certain DL cofolding methods are highly sensitive to their input multiple sequence alignments, whereas others are not; and (3) DL methods struggle to strike a balance between structural accuracy and chemical specificity when predicting new or multiligand protein targets. Morehead et al. introduce the benchmark PoseBench and evaluate the strengths and limitations of current AI-based protein–ligand docking and structure prediction methods.","PeriodicalId":48533,"journal":{"name":"Nature Machine Intelligence","volume":"8 1","pages":"32-41"},"PeriodicalIF":23.9,"publicationDate":"2025-12-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.nature.comhttps://www.nature.com/articles/s42256-025-01160-1.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145894622","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}
Pub Date : 2025-12-31DOI: 10.1038/s42256-025-01167-8
Chun-Teh Chen
Deep generative models that learn intermediate surface-current maps, rather than layouts directly, offer a more stable route to inverse design of tunable and stacked metasurfaces.
深度生成模型学习中间表面电流图，而不是直接布局，为可调和堆叠元表面的反向设计提供了更稳定的途径。
{"title":"Learning intermediate physical states for inverse metasurface design","authors":"Chun-Teh Chen","doi":"10.1038/s42256-025-01167-8","DOIUrl":"10.1038/s42256-025-01167-8","url":null,"abstract":"Deep generative models that learn intermediate surface-current maps, rather than layouts directly, offer a more stable route to inverse design of tunable and stacked metasurfaces.","PeriodicalId":48533,"journal":{"name":"Nature Machine Intelligence","volume":"8 1","pages":"2-3"},"PeriodicalIF":23.9,"publicationDate":"2025-12-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145895501","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}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：