检索结果-内蒙古大学图书馆

SpookyNet: learning force fields with electronic degrees of freedom and nonlocal effects

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Unke, Oliver T. Chmiela, Stefan Gastegger, Michael Schütt, Kristof T. Sauceda, Huziel E. Müller, Klaus-Robert Machine Learning Group Technische Universität Berlin Berlin10587 Germany Technische Universität Berlin Berlin10623 Germany BASLEARN BASF-TU joint Lab Technische Universität Berlin Berlin10587 Germany Department of Artificial Intelligence Korea University Anam-dong Seongbuk-gu Seoul02841 Korea Republic of Max Planck Institute for Informatics Stuhlsatzenhausweg Saarbrücken66123 Germany BIFOLD - Berlin Institute for the Foundations of Learning and Data Berlin Germany Google Research Brain team Berlin Germany

machine-learned force fields (ML-FFs) combine the accuracy of ab initio methods with the efficiency of conventional force fields. However, current ML-FFs typically ignore electronic degrees of freedom, such as the total charge or spin state, and assume chemical locality, which is problematic when molecules have inconsistent electronic states, or when nonlocal effects play a significant role. This work introduces SpookyNet, a deep neural network for constructing ML-FFs with explicit treatment of electronic degrees of freedom and quantum nonlocality. Chemically meaningful inductive biases and analytical corrections built into the network architecture allow it to properly model physical limits. SpookyNet improves upon the current state-of-the-art (or achieves similar performance) on popular quantum chemistry data sets. Notably, it is able to generalize across chemical and conformational space and can leverage the learned chemical insights, e.g. by predicting unknown spin states, thus helping to close a further important remaining gap for today's machine learning models in quantum chemistry. Copyright © 2021, The Authors. All rights reserved.

关键词： Spin dynamics

SE(3)-equivariant prediction of molecular wavefunctions and electronic densities

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Unke, Oliver T. Bogojeski, Mihail Gastegger, Michael Geiger, Mario Smidt, Tess Müller, Klaus-Robert Machine Learning Group Technische Universität Berlin Berlin10587 Germany Technische Universität Berlin Berlin10623 Germany Institute of Physics École Polytechnique Fédérale de Lausanne Lausanne1015 Switzerland Computational Research Division Lawrence Berkeley National Laboratory BerkeleyCA94720 United States Lawrence Berkeley National Laboratory BerkeleyCA94720 United States Department of Artificial Intelligence Korea University Anam-dong Seongbuk-gu Seoul02841 Korea Republic of Max Planck Institute for Informatics Stuhlsatzenhausweg Saarbrücken66123 Germany BIFOLD - Berlin Institute for the Foundations of Learning and Data Berlin Germany BASLEARN - TU Berlin BASF Joint Lab for Machine Learning Technische Universität Berlin Berlin10587 Germany Google Research Brain Team Berlin Germany

machine learning has enabled the prediction of quantum chemical properties with high accuracy and efficiency, allowing to bypass computationally costly ab initio calculations. Instead of training on a fixed set of properties, more recent approaches attempt to learn the electronic wavefunction (or density) as a central quantity of atomistic systems, from which all other observables can be derived. This is complicated by the fact that wavefunctions transform non-trivially under molecular rotations, which makes them a challenging prediction target. To solve this issue, we introduce general SE(3)-equivariant operations and building blocks for constructing deep learning architectures for geometric point cloud data and apply them to reconstruct wavefunctions of atomistic systems with unprecedented accuracy. Our model achieves speedups of over three orders of magnitude compared to ab initio methods and reduces prediction errors by up to two orders of magnitude compared to the previous state-of-the-art. This accuracy makes it possible to derive properties such as energies and forces directly from the wavefunction in an end-to-end manner. We demonstrate the potential of our approach in a transfer learning application, where a model trained on low accuracy reference wavefunctions implicitly learns to correct for electronic many-body interactions from observables computed at a higher level of theory. Such machine-learned wavefunction surrogates pave the way towards novel semi-empirical methods, offering resolution at an electronic level while drastically decreasing computational cost. Additionally, the predicted wavefunctions can serve as initial guess in conventional ab initio methods, decreasing the number of iterations required to arrive at a converged solution, thus leading to significant speedups without any loss of accuracy or robustness. While we focus on physics applications in this contribution, the proposed equivariant framework for deep learning on point clouds is p

关键词： Stereochemistry

LEVERAGING WEAK COMPLEMENTARY LABELS TO IMPROVE SEMANTIC SEGMENTATION OF HEPATOCELLULAR CARCINOMA AND CHOLANGIOCARCINOMA IN H&E-STAINED SLIDES

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Hägele, Miriam Eschrich, Johannes Schallenberg, Simon Guillot, Adrien Ruff, Lukas Roderburg, Christoph Alber, Maximilian Tacke, Frank Klauschen, Frederick Machine Learning Group Technische Universität Berlin Germany BIFOLD Berlin Institute for the Foundations of Learning and Data Germany Aignostics GmbH Germany Department of Hepatology and Gastroenterology Charité Universitätsmedizin Berlin Germany Berlin Institute of Health at Charité Universitätsmedizin Berlin Germany Institute of Pathology Charité Universitätsmedizin Berlin Germany Clinic for Gastroenterology Hepatology and Infectious Diseases University Hospital Düsseldorf Germany Institute of Pathology Ludwig-Maximilians-Universität München Germany German Cancer Consortium Munich Partner Site German Cancer Research Center Germany

In this paper, we present a deep learning segmentation approach to classify and quantify the two most prevalent primary liver cancers – hepatocellular carcinoma and intrahepatic cholangiocarcinoma – from hematoxylin and eosin (H&E) stained whole slide images. While semantic segmentation of medical images typically requires costly pixel-level annotations by domain experts, there often exists additional information which is routinely obtained in clinical diagnostics but rarely utilized for model training. We propose to leverage such weak information from patient diagnoses by deriving complementary labels that indicate to which class a sample cannot belong to. To integrate these labels, we formulate a complementary loss for segmentation. Motivated by the medical application, we demonstrate for general segmentation tasks that including additional patches with solely weak complementary labels during model training can significantly improve the predictive performance and robustness of a model. On the task of diagnostic differentiation between hepatocellular carcinoma and intrahepatic cholangiocarcinoma, we achieve a balanced accuracy of 0.91 (CI 95%: 0.86 − 0.95) at case level for 165 hold-out patients. Furthermore, we also show that leveraging complementary labels improves the robustness of segmentation and increases performance at case level. © 2023, CC BY.

关键词： Medical applications

Traffic4cast at NeurIPS 2021 – Temporal and Spatial Few-Shot Transfer learning in Gridded Geo-Spatial Processes

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Eichenberger, Christian Neun, Moritz Martin, Henry Herruzo, Pedro Spanring, Markus Lu, Yichao Choi, Sungbin Konyakhin, Vsevolod Lukashina, Nina Shpilman, Aleksei Wiedemann, Nina Raubal, Martin Wang, Bo Vu, Hai L. Mohajerpoor, Reza Cai, Chen Kim, Inhi Hermes, Luca Melnik, Andrew Velioglu, Riza Vieth, Markus Schilling, Malte Bojesomo, Alabi Al Marzouqi, Hasan Liatsis, Panos Santokhi, Jay Hillier, Dylan Yang, Yiming Sarwar, Joned Jordan, Anna Hewage, Emil Jonietz, David Tang, Fei Gruca, Aleksandra Kopp, Michael Kreil, David Hochreiter, Sepp Vienna Austria Institute of Cartography and Geoinformation ETH Zurich Switzerland Layer 6 AI Toronto Canada ITMO University Saint Petersburg Russia JetBrains Research Saint Petersburg Russia HSE University Saint Petersburg Russia Institute of Transport Studies Monash University ClaytonVIC Australia CSIRO’s Data61 Eveleigh Australia Institute Civil and Environmental Engineering Department Kongju National University Korea Republic of Machine Learning & Neuroinformatics Group Bielefeld University Germany Electrical Engineering and Computer Science Department Khalifa University Abu Dhabi United Arab Emirates Alchera Data Technologies Ltd Cambridge United Kingdom HERE Technologies Zurich Switzerland Silesian University of Technology Gliwice Poland Machine Learning Institute Johannes Kepler University Linz Austria

The IARAI Traffic4cast competitions at NeurIPS 2019 and 2020 showed that neural networks can successfully predict future traffic conditions 1 hour into the future on simply aggregated GPS probe data in time and space bins. We thus reinterpreted the challenge of forecasting traffic conditions as a movie completion task. U-Nets proved to be the winning architecture, demonstrating an ability to extract relevant features in this complex real-world geo-spatial process. Building on the previous competitions, Traffic4cast 2021 now focuses on the question of model robustness and generalizability across time and space. Moving from one city to an entirely different city, or moving from pre-COVID times to times after COVID hit the world thus introduces a clear domain shift. We thus, for the first time, release data featuring such domain shifts. The competition now covers ten cities over 2 years, providing data compiled from over 1012 GPS probe data. Winning solutions captured traffic dynamics sufficiently well to even cope with these complex domain shifts. Surprisingly, this seemed to require only the previous 1h traffic dynamic history and static road graph as input. © 2022, CC BY-NC-SA.

关键词： Probes

Accurate global machine learning force fields for molecules with hundreds of atoms

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Chmiela, Stefan Vassilev-Galindo, Valentin Unke, Oliver T. Kabylda, Adil Sauceda, Huziel E. Tkatchenko, Alexandre Müller, Klaus-Robert Machine Learning Group Technische Universität Berlin Berlin10587 Germany Berlin Institute for the Foundations of Learning and Data – BIFOLD Germany Department of Physics and Materials Science University of Luxembourg Luxembourg CityL-1511 Luxembourg Google Research Brain Team Berlin Germany Departamento de Materia Condensada Instituto de Física Universidad Nacional Autónoma de México Cd. de MéxicoC.P. 04510 Mexico BASLEARN - TU Berlin BASF Joint Lab for Machine Learning Technische Universität Berlin Berlin10587 Germany Max Planck Institute for Informatics Stuhlsatzenhausweg Saarbrücken66123 Germany Department of Artificial Intelligence Korea University Anam-dong Seongbuk-gu Seoul02841 Korea Republic of

Global machine learning force fields (MLFFs), that have the capacity to capture collective many-atom interactions in molecular systems, currently only scale up to a few dozen atoms due a considerable growth of the model complexity with system size. For larger molecules, locality assumptions are typically introduced, with the consequence that non-local interactions are poorly or not at all described, even if those interactions are contained within the reference ab initio data. Here, we approach this challenge and develop an exact iterative parameter-free approach to train global symmetric gradient domain machine learning (sGDML) force fields for systems with up to several hundred atoms, without resorting to any localization of atomic interactions or other potentially uncontrolled approximations. This means that all atomic degrees of freedom remain fully correlated in the global sGDML FF, allowing the accurate description of complex molecules and materials that present phenomena with far-reaching characteristic correlation lengths. We assess the accuracy and efficiency of our MLFFs on a newly developed MD22 benchmark dataset containing molecules from 42 to 370 atoms. The robustness of our approach is demonstrated in nanosecond long path-integral molecular dynamics simulations for the supramolecular complexes in the MD22 dataset. Copyright © 2022, The Authors. All rights reserved.

关键词： machine learning

Strategic priorities for transformative progress in advancing biology with proteomics and artificial intelligence

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Sun, Yingying Jun, A. Liu, Zhiwei Sun, Rui Qian, Liujia Payne, Samuel H. Bittremieux, Wout Ralser, Markus Li, Chen Chen, Yi Dong, Zhen Perez-Riverol, Yasset Khan, Asif Sander, Chris Aebersold, Ruedi Vizcaíno, Juan Antonio Krieger, Jonathan R. Yao, Jianhua Wen, Han Zhang, Linfeng Zhu, Yunping Xuan, Yue Sun, Benjamin Boyang Qiao, Liang Hermjakob, Henning Tang, Haixu Gao, Huanhuan Deng, Yamin Zhong, Qing Chang, Cheng Bandeira, Nuno Li, Ming Weinan, E. Sun, Siqi Yang, Yuedong Omenn, Gilbert S. Zhang, Yue Xu, Ping Fu, Yan Liu, Xiaowen Overall, Christopher M. Wang, Yu Deutsch, Eric W. Chen, Luonan Cox, Jürgen Demichev, Vadim He, Fuchu Huang, Jiaxing Jin, Huilin Liu, Chao Li, Nan Luan, Zhongzhi Song, Jiangning Yu, Kaicheng Wan, Wanggen Wang, Tai Zhang, Kang Zhang, Le Bell, Peter A. Mann, Matthias Zhang, Bing Guo, Tiannan Affiliated Hangzhou First People’s Hospital State Key Laboratory of Medical Proteomics School of Medicine Westlake University Zhejiang Province Hangzhou China Westlake Center for Intelligent Proteomics Westlake Laboratory of Life Sciences and Biomedicine Zhejiang Province Hangzhou China Biology Department Brigham Young University ProvoUT84602 United States Department of Computer Science University of Antwerp Antwerp2020 Belgium Department of Biochemistry CharitéUniversitätsmedizin Berlin Berlin Germany Biomedicine Discovery Institute Department of Biochemistry and Molecular Biology Monash University MelbourneVICVIC 3800 Australia Wellcome Genome Campus Hinxton CambridgeCB10 1SD United Kingdom Harvard Medical School Ludwig Center at Harvard United States Harvard Medical School Broad Institute Ludwig Center at Harvard Dana-Farber Cancer Institute United States Department of Biology Institute of Molecular Systems Biology ETH Zürich Zürich Switzerland Bruker Ltd. MiltonONL9T 6P4 Canada AI for Life Sciences Lab Tencent Shenzhen518057 China State Key Laboratory of Medical Proteomics AI for Science Institute Beijing100080 China Beijing Institute of Lifeomics Beijing102206 China Thermo Fisher Scientific GmbH Hanna-Kunath Str. 11 Bremen28199 Germany Informatics and Predictive Sciences Research Bristol Myers Squibb United States Department of Chemistry Fudan University Songhu Road 2005 Shanghai200438 China Department of Computer Science Luddy School of Informatics Computing and Engineering Indiana University IN47408 United States ProCan® Children’s Medical Research Institute Faculty of Medicine and Health The University of Sydney WestmeadNSW Australia La Jolla CA United States Central China Institute of Artificial Intelligence University of Waterloo Canada AI for Science Institute Center for Machine Learning Research School of Mathematical Sciences Peking University China Research Institute of Intelligent Complex Systems Fudan U

Artificial intelligence (AI) is transforming scientific research, including proteomics. Advances in mass spectrometry (MS)-based proteomics data quality, diversity, and scale, combined with groundbreaking AI techniques, are unlocking new challenges and opportunities in biological discovery. Here, we highlight key areas where AI is driving innovation, from data analysis to new biological insights. These include developing an AI-friendly ecosystem for proteomics data generation, sharing, and analysis;improving peptide and protein identification and quantification;characterizing protein-protein interactions and protein complexes;advancing spatial and perturbation proteomics;integrating multi-omics data;and ultimately enabling AI-empowered virtual cells. © 2025, CC BY.

关键词： Biotic

Evaluating deep transfer learning for whole-brain cognitive decoding

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Thomas, Armin W. Lindenberger, Ulman Samek, Wojciech Müller, Klaus-Robert Machine Learning Group Dept. of Computer Science and Electrical Engineering Technische Universität Berlin Berlin Germany Center for Lifespan Psychology Max Planck Institute for Human Development Berlin Germany Stanford Data Science Stanford University StanfordCA United States Dept. of Psychology Stanford University StanfordCA United States Max Planck UCL Centre for Computational Psychiatry and Ageing Research Berlin Germany Dept. of Artificial Intelligence Fraunhofer Heinrich Hertz Institute Berlin Germany BIFOLD – Berlin Institute for the Foundations of Learning and Data Berlin Germany Dept. of Artificial Intelligence Korea University Seoul Korea Republic of Max Planck Institute for Informatics Saarbrücken Germany

research in many fields has shown that transfer learning (TL) is well-suited to improve the performance of deep learning (DL) models in datasets with small numbers of samples. This empirical success has triggered interest in the application of TL to cognitive decoding analyses with functional neuroimaging data. Here, we systematically evaluate TL for the application of DL models to the decoding of cognitive states (e.g., viewing images of faces or houses) from whole-brain functional Magnetic Resonance Imaging (fMRI) data. We first pre-train two DL architectures on a large, public fMRI dataset and subsequently evaluate their performance in an independent experimental task and a fully independent dataset. The pre-trained models consistently achieve higher decoding accuracies and generally require less training time and data than model variants that were not pre-trained, clearly underlining the benefits of pre-training. We demonstrate that these benefits arise from the ability of the pre-trained models to reuse many of their learned features when training with new data, providing deeper insights into the mechanisms giving rise to the benefits of pre-training. Yet, we also surface nuanced challenges for whole-brain cognitive decoding with DL models when interpreting the decoding decisions of the pre-trained models, as these have learned to utilize the fMRI data in unforeseen and counterintuitive ways to identify individual cognitive states. © 2021, CC BY.

关键词： Decoding

BIGDML: Towards exact machine learning force fields for materials

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Sauceda, Huziel E. Gálvez-González, Luis E. Chmiela, Stefan Paz-Borbón, Lauro Oliver Müller, Klaus-Robert Tkatchenko, Alexandre Machine Learning Group Technische Universität Berlin Berlin10587 Germany BASLEARN TU Berlin BASF Joint Lab for Machine Learning Technische Universität Berlin Berlin10587 Germany División de Ciencias Exactas y Naturales Universidad de Sonora Blvd. Luis Encinas & Rosales Hermosillo Mexico BIFOLD – Berlin Institute for the Foundations of Learning and Data Germany Instituto de Física Universidad Nacional Autónoma de México Apartado Postal 20-364 CDMX01000 Mexico Google Research Brain team Berlin Germany Department of Artificial Intelligence Korea University Anam-dong Seongbuk-gu Seoul02841 Korea Republic of Max Planck Institute for Informatics Stuhlsatzenhausweg Saarbrücken66123 Germany Department of Physics and Materials Science University of Luxembourg LuxembourgL-1511 Luxembourg

machine-learning force fields (MLFF) should be accurate, computationally and data efficient, and applicable to molecules, materials, and interfaces thereof. Currently, MLFFs often introduce tradeoffs that restrict their practical applicability to small subsets of chemical space or require exhaustive datasets for training. Here, we introduce the Bravais-Inspired Gradient-Domain machine learning (BIGDML) approach and demonstrate its ability to construct reliable force fields using a training set with just 10-200 geometries for materials including pristine and defect-containing 2D and 3D semiconductors and metals, as well as chemisorbed and physisorbed atomic and molecular adsorbates on surfaces. The BIGDML model employs the full relevant symmetry group for a given material, does not assume artificial atom types or localization of atomic interactions and exhibits high data efficiency and state-of-the-art energy accuracies (errors substantially below 1 meV per atom) for an extended set of materials. Extensive path-integral molecular dynamics carried out with BIGDML models demonstrate the counterintuitive localization of benzene–graphene dynamics induced by nuclear quantum effects and allow to rationalize the Arrhenius behavior of hydrogen diffusion coefficient in a Pd crystal for a wide range of temperatures. Copyright © 2021, The Authors. All rights reserved.

关键词： machine learning

Scholarly Influence of the Conference and Labs of the Evaluation Forum eHealth Initiative: Review and Bibliometric Study of the 2012 to 2017 Outcomes

学校读者我要写书评

暂无评论

JMIR research Protocols 2018年第7期7卷 e10961页

作者： Suominen, Hanna Kelly, Liadh Goeuriot, Lorraine Research School of Computer Science College of Engineering and Computer Science The Australian National University Canberra ACT Australia Machine Learning Research Group Data61 Commonwealth Scientific and Industrial Research Organisation Canberra ACT Australia Faculty of Science and Technology University of Canberra Canberra ACT Australia Department of Future Technologies Faculty of Science and Engineering University of Turku Turku Finland Department of Computer Science Maynooth University Maynooth Co Kildare Ireland Grenoble Informatics Laboratory Université Grenoble Alpes Grenoble France

Background: The eHealth initiative of the Conference and Labs of the Evaluation Forum (CLEF) has aimed since 2012 to provide researchers working on health text analytics with annual workshops, shared development challenges and tasks, benchmark datasets, and software for processing and evaluation. In 2012, it ran as a scientific workshop with the aim of establishing an evaluation lab, and since 2013, this annual workshop has been supplemented with 3 or more preceding labs each year. An evaluation lab is an activity where the participating individuals or teams’ goal is to solve the same problem, typically using the same dataset in a given time frame. The overall purpose of this initiative is to support patients, their next of kin, clinical staff, health scientists, and health care policy makers in accessing, understanding, using, and authoring health information in a multilingual setting. In the CLEF eHealth 2013 to 2017 installations, the aim was to address patient-centric text processing. From 2015, the scope was also extended to aid both patients’ understanding and clinicians’ authoring of various types of medical content. CLEF eHealth 2017 introduced a new pilot task on technology-assisted reviews (TARs) in empirical medicine in order to support health scientists and health care policymakers’ information access. Objectives: This original research paper reports on the outcomes of the first 6 installations of CLEF eHealth from 2012 to 2017. The focus is on measuring and analyzing the scholarly influence by reviewing CLEF eHealth papers and their citations. Methods: A review and bibliometric study of the CLEF eHealth proceedings, working notes, and author-declared paper extensions were conducted. Citation content analysis was used for the publications and their citations collected from Google Scholar. Results: As many as 718 teams registered their interest in the tasks, leading to 130 teams submitting to the 15 tasks. A total of 184 papers using CLEF eHealth data gen

关键词： Evaluation studies as topic Health records Information extraction Information storage and retrieval Information visualization Patient education as topic Speech recognition Systematic reviews Test-set generation Text classification