Nature Machine Intelligence最新文献

{"title":"Rethinking machine unlearning for large language models","authors":"Sijia Liu, Yuanshun Yao, Jinghan Jia, Stephen Casper, Nathalie Baracaldo, Peter Hase, Yuguang Yao, Chris Yuhao Liu, Xiaojun Xu, Hang Li, Kush R. Varshney, Mohit Bansal, Sanmi Koyejo, Yang Liu","doi":"10.1038/s42256-025-00985-0","DOIUrl":"10.1038/s42256-025-00985-0","url":null,"abstract":"We explore machine unlearning in the domain of large language models (LLMs), referred to as LLM unlearning. This initiative aims to eliminate undesirable data influence (for example, sensitive or illegal information) and the associated model capabilities, while maintaining the integrity of essential knowledge generation and not affecting causally unrelated information. We envision LLM unlearning becoming a pivotal element in the life-cycle management of LLMs, potentially standing as an essential foundation for developing generative artificial intelligence that is not only safe, secure and trustworthy but also resource-efficient without the need for full retraining. We navigate the unlearning landscape in LLMs from conceptual formulation, methodologies, metrics and applications. In particular, we highlight the often-overlooked aspects of existing LLM unlearning research, for example, unlearning scope, data–model interaction and multifaceted efficacy assessment. We also draw connections between LLM unlearning and related areas such as model editing, influence functions, model explanation, adversarial training and reinforcement learning. Furthermore, we outline an effective assessment framework for LLM unlearning and explore its applications in copyright and privacy safeguards and sociotechnical harm reduction. Machine unlearning techniques remove undesirable data and associated model capabilities while preserving essential knowledge, so that machine learning models can be updated without costly retraining. Liu et al. review recent advances and opportunities in machine unlearning in LLMs, revisiting methodologies and overlooked principles for future improvements and exploring emerging applications in copyright and privacy safeguards and in reducing sociotechnical harms.","PeriodicalId":48533,"journal":{"name":"Nature Machine Intelligence","volume":"7 2","pages":"181-194"},"PeriodicalIF":18.8,"publicationDate":"2025-02-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143427029","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}

{"title":"Towards a more inductive world for drug repurposing approaches","authors":"Jesus de la Fuente, Guillermo Serrano, Uxía Veleiro, Mikel Casals, Laura Vera, Marija Pizurica, Nuria Gómez-Cebrián, Leonor Puchades-Carrasco, Antonio Pineda-Lucena, Idoia Ochoa, Silve Vicent, Olivier Gevaert, Mikel Hernaez","doi":"10.1038/s42256-025-00987-y","DOIUrl":"10.1038/s42256-025-00987-y","url":null,"abstract":"Drug–target interaction (DTI) prediction is a challenging albeit essential task in drug repurposing. Learning on graph models has drawn special attention as they can substantially reduce drug repurposing costs and time commitment. However, many current approaches require high-demand additional information besides DTIs that complicates their evaluation process and usability. Additionally, structural differences in the learning architecture of current models hinder their fair benchmarking. In this work, we first perform an in-depth evaluation of current DTI datasets and prediction models through a robust benchmarking process and show that DTI methods based on transductive models lack generalization and lead to inflated performance when traditionally evaluated, making them unsuitable for drug repurposing. We then propose a biologically driven strategy for negative-edge subsampling and uncovered previously unknown interactions via in vitro validation, missed by traditional subsampling. Finally, we provide a toolbox from all generated resources, crucial for fair benchmarking and robust model design. The authors address the challenge of predicting drug–target interactions, which is crucial for drug repurposing, by introducing a robust benchmarking framework. Using a biologically driven strategy, they uncover previously unknown interactions.","PeriodicalId":48533,"journal":{"name":"Nature Machine Intelligence","volume":"7 3","pages":"495-508"},"PeriodicalIF":18.8,"publicationDate":"2025-02-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.nature.com/articles/s42256-025-00987-y.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143401247","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}

{"title":"Benchmarking AI-powered docking methods from the perspective of virtual screening","authors":"Shukai Gu, Chao Shen, Xujun Zhang, Huiyong Sun, Heng Cai, Hao Luo, Huifeng Zhao, Bo Liu, Hongyan Du, Yihao Zhao, Chenggong Fu, Silong Zhai, Yafeng Deng, Huanxiang Liu, Tingjun Hou, Yu Kang","doi":"10.1038/s42256-025-00993-0","DOIUrl":"10.1038/s42256-025-00993-0","url":null,"abstract":"Recently, many artificial intelligence (AI)-powered protein–ligand docking and scoring methods have been developed, demonstrating impressive speed and accuracy. However, these methods often neglected the physical plausibility of the docked complexes and their efficacy in virtual screening (VS) projects. Therefore, we conducted a comprehensive benchmark analysis of four AI-powered and four physics-based docking tools and two AI-enhanced rescoring methods. We initially constructed the TrueDecoy set, a dataset on which the redocking experiments revealed that KarmaDock and CarsiDock surpassed all physics-based tools in docking accuracy, whereas all physics-based tools notably outperformed AI-based methods in structural rationality. The low physical plausibility of docked structures generated by the top AI method, CarsiDock, mainly stems from insufficient intermolecular validity. The VS results on the TrueDecoy set highlight the effectiveness of RTMScore as a rescore function, and Glide-based methods achieved the highest enrichment factors among all docking tools. Furthermore, we created the RandomDecoy set, a dataset that more closely resembles real-world VS scenarios, where AI-based tools obviously outperformed Glide. Additionally, we found that the employed ligand-based postprocessing methods had a weak or even negative impact on optimizing the conformations of docked complexes and enhancing VS performance. Finally, we proposed a hierarchical VS strategy that could efficiently and accurately enrich active molecules in large-scale VS projects. Artificial intelligence (AI)-based docking and scoring methods demonstrate considerable potential for virtual drug screening. Gu et al. go further by assessing the structural rationality of AI-predicted complex conformations from various sources.","PeriodicalId":48533,"journal":{"name":"Nature Machine Intelligence","volume":"7 3","pages":"509-520"},"PeriodicalIF":18.8,"publicationDate":"2025-02-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143401248","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}

{"title":"Image-based generation for molecule design with SketchMol","authors":"Zixu Wang, Yangyang Chen, Pengsen Ma, Zhou Yu, Jianmin Wang, Yuansheng Liu, Xiucai Ye, Tetsuya Sakurai, Xiangxiang Zeng","doi":"10.1038/s42256-025-00982-3","DOIUrl":"10.1038/s42256-025-00982-3","url":null,"abstract":"Efficient molecular design methods are crucial for accelerating early stage drug discovery, potentially saving years of development time and billions of dollars in costs. Current molecular design methods rely on sequence-based or graph-based representations, emphasizing local features such as bonds and atoms but lacking a comprehensive depiction of the overall molecular topology. Here we introduce SketchMol, an image-based molecular generation framework that combines visual understanding with molecular design. SketchMol leverages diffusion models and applies a refinement technique called reinforcement learning from molecular experts to improve the generation of viable molecules. It creates molecules through a painting-like approach that simultaneously depicts local structures and global layout of the molecule. By visualizing molecular structures, various design tasks are unified within a single image-based framework. De novo design becomes sketching new molecular images, whereas editing tasks transform into filling partially drawn images. Through extensive experiments, we demonstrated that SketchMol effectively handles a variety of molecular design tasks. SketchMol is a model that explores the feasibility of incorporating image generation techniques into the field of small-molecule design.","PeriodicalId":48533,"journal":{"name":"Nature Machine Intelligence","volume":"7 2","pages":"244-255"},"PeriodicalIF":18.8,"publicationDate":"2025-02-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143401246","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}

{"title":"On board with COMET to improve omics prediction models","authors":"Paul Fogel, George Luta","doi":"10.1038/s42256-025-00990-3","DOIUrl":"10.1038/s42256-025-00990-3","url":null,"abstract":"The performance of omics prediction models can be significantly improved by combining limited patient proteomic data with widely available electronic health records.","PeriodicalId":48533,"journal":{"name":"Nature Machine Intelligence","volume":"7 2","pages":"168-169"},"PeriodicalIF":18.8,"publicationDate":"2025-02-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143393225","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}

{"title":"On the caveats of AI autophagy","authors":"Xiaodan Xing, Fadong Shi, Jiahao Huang, Yinzhe Wu, Yang Nan, Sheng Zhang, Yingying Fang, Michael Roberts, Carola-Bibiane Schönlieb, Javier Del Ser, Guang Yang","doi":"10.1038/s42256-025-00984-1","DOIUrl":"10.1038/s42256-025-00984-1","url":null,"abstract":"Generative artificial intelligence (AI) technologies and large models are producing realistic outputs across various domains, such as images, text, speech and music. Creating these advanced generative models requires significant resources, particularly large and high-quality datasets. To minimize training expenses, many algorithm developers use data created by the models themselves as a cost-effective training solution. However, not all synthetic data effectively improve model performance, necessitating a strategic balance in the use of real versus synthetic data to optimize outcomes. Currently, the previously well-controlled integration of real and synthetic data is becoming uncontrollable. The widespread and unregulated dissemination of synthetic data online leads to the contamination of datasets traditionally compiled through web scraping, now mixed with unlabelled synthetic data. This trend, known as the AI autophagy phenomenon, suggests a future where generative AI systems may increasingly consume their own outputs without discernment, raising concerns about model performance, reliability and ethical implications. What will happen if generative AI continuously consumes itself without discernment? What measures can we take to mitigate the potential adverse effects? To address these research questions, this Perspective examines the existing literature, delving into the consequences of AI autophagy, analysing the associated risks and exploring strategies to mitigate its impact. Our aim is to provide a comprehensive perspective on this phenomenon advocating for a balanced approach that promotes the sustainable development of generative AI technologies in the era of large models. With widespread generation and availability of synthetic data, AI systems are increasingly trained on their own outputs, leading to various technical and ethical challenges. The authors analyse this development and discuss measures to mitigate the potential adverse effects of ‘AI eating itself’.","PeriodicalId":48533,"journal":{"name":"Nature Machine Intelligence","volume":"7 2","pages":"172-180"},"PeriodicalIF":18.8,"publicationDate":"2025-02-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143375174","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}

{"title":"The promise of generative AI for suicide prevention in India","authors":"Tanmoy Chakraborty, Koushik Sinha Deb, Himanshu Kulkarni, Sarah Masud, Suresh Bada Math, Gayatri Oke, Rajesh Sagar, Mona Sharma","doi":"10.1038/s42256-025-00992-1","DOIUrl":"10.1038/s42256-025-00992-1","url":null,"abstract":"","PeriodicalId":48533,"journal":{"name":"Nature Machine Intelligence","volume":"7 2","pages":"162-163"},"PeriodicalIF":18.8,"publicationDate":"2025-02-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143191830","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}

{"title":"Discovering fully semantic representations via centroid- and orientation-aware feature learning","authors":"Jaehoon Cha, Jinhae Park, Samuel Pinilla, Kyle L. Morris, Christopher S. Allen, Mark I. Wilkinson, Jeyan Thiyagalingam","doi":"10.1038/s42256-024-00978-5","DOIUrl":"10.1038/s42256-024-00978-5","url":null,"abstract":"Learning meaningful representations of images in scientific domains that are robust to variations in centroids and orientations remains an important challenge. Here we introduce centroid- and orientation-aware disentangling autoencoder (CODAE), an encoder–decoder-based neural network that learns meaningful content of objects in a latent space. Specifically, a combination of a translation- and rotation-equivariant encoder, Euler encoding and an image moment loss enables CODAE to extract features invariant to positions and orientations of objects of interest from randomly translated and rotated images. We evaluate this approach on several publicly available scientific datasets, including protein images from life sciences, four-dimensional scanning transmission electron microscopy data from material science and galaxy images from astronomy. The evaluation shows that CODAE learns centroids, orientations and their invariant features and outputs, as well as aligned reconstructions and the exact view reconstructions of the input images with high quality. Cha and colleagues present a translation- and rotation-equivariant autoencoder-based method for robust image recognition, which they demonstrate on diverse tasks from bioinformatics, material science and astronomy.","PeriodicalId":48533,"journal":{"name":"Nature Machine Intelligence","volume":"7 2","pages":"307-314"},"PeriodicalIF":18.8,"publicationDate":"2025-02-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.nature.com/articles/s42256-024-00978-5.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143191831","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}

{"title":"Preserving and combining knowledge in robotic lifelong reinforcement learning","authors":"Yuan Meng, Zhenshan Bing, Xiangtong Yao, Kejia Chen, Kai Huang, Yang Gao, Fuchun Sun, Alois Knoll","doi":"10.1038/s42256-025-00983-2","DOIUrl":"10.1038/s42256-025-00983-2","url":null,"abstract":"Humans can continually accumulate knowledge and develop increasingly complex behaviours and skills throughout their lives, which is a capability known as ‘lifelong learning’. Although this lifelong learning capability is considered an essential mechanism that makes up general intelligence, recent advancements in artificial intelligence predominantly excel in narrow, specialized domains and generally lack this lifelong learning capability. Here we introduce a robotic lifelong reinforcement learning framework that addresses this gap by developing a knowledge space inspired by the Bayesian non-parametric domain. In addition, we enhance the agent’s semantic understanding of tasks by integrating language embeddings into the framework. Our proposed embodied agent can consistently accumulate knowledge from a continuous stream of one-time feeding tasks. Furthermore, our agent can tackle challenging real-world long-horizon tasks by combining and reapplying its acquired knowledge from the original tasks stream. The proposed framework advances our understanding of the robotic lifelong learning process and may inspire the development of more broadly applicable intelligence. Humans continuously acquire knowledge and develop complex behaviours. Meng, Bing, Yao and colleagues present a robotic lifelong learning framework using a Bayesian non-parametric knowledge space, enabling agents to dynamically preserve and integrate knowledge from sequential tasks, enhancing adaptability.","PeriodicalId":48533,"journal":{"name":"Nature Machine Intelligence","volume":"7 2","pages":"256-269"},"PeriodicalIF":18.8,"publicationDate":"2025-02-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.nature.com/articles/s42256-025-00983-2.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143125412","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}

{"title":"Why the carbon footprint of generative large language models alone will not help us assess their sustainability","authors":"Leonie N. Bossert, Wulf Loh","doi":"10.1038/s42256-025-00979-y","DOIUrl":"10.1038/s42256-025-00979-y","url":null,"abstract":"There is a growing awareness of the substantial environmental costs of large language models (LLMs), but discussing the sustainability of LLMs only in terms of CO2 emissions is not enough. This Comment emphasizes the need to take into account the social and ecological costs and benefits of LLMs as well.","PeriodicalId":48533,"journal":{"name":"Nature Machine Intelligence","volume":"7 2","pages":"164-165"},"PeriodicalIF":18.8,"publicationDate":"2025-02-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143077564","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}