Machine Learning: Science and Technology最新文献

{"title":"Transferring predictions of formation energy across lattices of increasing size","authors":"Massimiliano Lupo Pasini, Mariia Karabin, M. Eisenbach","doi":"10.1088/2632-2153/ad3d2c","DOIUrl":"https://doi.org/10.1088/2632-2153/ad3d2c","url":null,"abstract":"\u0000 In this study, we show the transferability of graph convolutional neural network (GCNN) predictions of the formation energy of the nickel-platinum (NiPt) solid solution alloy across atomic structures of increasing sizes. The original dataset was generated with the Large-scale Atomic/Molecular Massively Parallel Simulator (LAMMPS) using the second nearest-neighbor modified embedded-atom method (2NN MEAM) empirical interatomic potential. Geometry optimization was performed on the initially randomly generated face centered cubic (FCC) crystal structures and the formation energy has been calculated at each step of the geometry optimization, with configurations spanning the whole compositional range. Using data from various steps of the geometry optimization, we first trained our open-source, scalable implementation of GCNN called HydraGNN on a lattice of 256 atoms, which accounts well for the short-range interactions. Using this data, we predicted the formation energy for lattices of 864 atoms and 2,048 atoms, which resulted in lower-than-expected accuracy due to the long-range interactions present in these larger lattices. We accounted for the long-range interactions by including a small amount of training data representative for those two larger sizes, whereupon the predictions of HydraGNN scaled linearly with the size of the lattice. Therefore, our strategy ensured scalability while reducing significantly the computational cost of training on larger lattice sizes.","PeriodicalId":503691,"journal":{"name":"Machine Learning: Science and Technology","volume":"62 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-04-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140716931","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Neural Networks for Separation of Cosmic Gamma Rays and Hadronic Cosmic Rays in Air Shower Observation with a Large Area Surface Detector Array","authors":"S. Okukawa, Kazuyuki Hara, K. Hibino, Y. Katayose, K. Kawata, M. Ohnishi, Takashi Sako, Takashi Sako, Makio Shibata, A. Shiomi, M. Takita","doi":"10.1088/2632-2153/ad3a33","DOIUrl":"https://doi.org/10.1088/2632-2153/ad3a33","url":null,"abstract":"\u0000 The Tibet ASγ experiment has been observing cosmic gamma rays and cosmic rays in the energy range from teraelectron volts to several tens of petaelectron volts with a surface detector array since 1990. The derivation of cosmic gamma-ray flux is made by finding the excess distribution of the arrival direction of air showers above background cosmic rays. In 2014, the underground water Cherenkov muon detector (MD) was added to separate cosmic gamma rays from the background on the basis of the muon-less feature of the air showers of gamma-ray origin; hybrid observations using these two detectors were started at this time. In the present study, we developed methods to separate gamma-ray-induced air showers and hadronic cosmic-ray-induced ones using the measured particle number density distribution to improve the sensitivity of cosmic gamma-ray measurements using the Tibet air shower array data alone before the installation of the MD. We tested two approaches based on neural networks. The first method used feature values representing the lateral spread of the secondary particles, and the second method used the shower image data. To compare the separation performance of each method, we analyzed Monte Carlo air shower events in the vertically incident direction with mono-initial-energy gamma rays and protons. When discriminated by a single feature, the feature with the highest separation performance has an area under the curve (AUC) value of 0.701 for a gamma-ray energy of 10 TeV and 0.808 for 100 TeV. A separation method with a multilayer perceptron (MLP) based on multiple features has AUC values of 0.761 for a gamma-ray energy of 10 TeV and 0.854 for 100 TeV, which represents an improvement of approximately 5 % in the AUC value compared with the single-feature case. We also found that the feature values that effectively contribute to the separation vary depending on the energy. A separation method with a convolutional neural network (CNN) using the shower image data has AUC values of 0.781 for a gamma-ray energy of 10 TeV and 0.901 for 100 TeV, which are approximately 5 % higher than those of the MLP method. We applied the CNN separation method to Monte Carlo gamma-ray and cosmic-ray events from the Crab Nebula in the energy range 10−100 TeV. The AUC values range from 0.753 to 0.879, and the significance of the observed gamma-ray excess is improved by 1.3 to 1.8 times compared with the case without the separation procedure.","PeriodicalId":503691,"journal":{"name":"Machine Learning: Science and Technology","volume":"12 2","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-04-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140747976","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Prediction of Mild Cognitive Impairment Using EEG Signal and BiLSTM Network","authors":"Tahani Jaser Alahmadi, Atta Ur Rahman, Zaid Ali Alhababi, Sania Ali, H. Alkahtani","doi":"10.1088/2632-2153/ad38fe","DOIUrl":"https://doi.org/10.1088/2632-2153/ad38fe","url":null,"abstract":"\u0000 Mild Cognitive Impairment (MCI) is a cognitive disease that primarily affects elderly persons. Patients with MCI have impairments in one or more cognitive areas, such as memory, attention, language, and problem-solving. The risk of Alzheimer's disease (AD) development is 10 times higher among individuals who meet the MCI diagnosis than in those who do not have such a diagnosis. Identifying the primary neurophysiological variations between those who are suffering from cognitive impairment and those who are ageing normally may provide helpful techniques to assess the effectiveness of therapies. Event-related Potentials (ERPs) are utilized to investigate the processing of sensory, cognitive, and motor information in the brain. ERPs enable excellent temporal resolution of underlying brain activity. ERP data is complex due to the temporal variation occurs in time domain. It is actually a type of electroencephalography (EEG) signal that is time-locked to a specific event or behavior. To remove artifacts from the data, this work utilizes Independent component analysis (ICA), finite impulse response (FIR) filter, and Fast Fourier Transformation (FFT) as preprocessing techniques. The Bidirectional Long Short-Term Memory (BiLSTM) network is utilized to retains the spatial relationships between the ERP data while learning changes in temporal information for a long time. This network performed well both in modeling and information extraction from the signals. To validate the model performance, the proposed framework is tested on two benchmark datasets. The proposed framework achieved state-of-the-art accuracy of 96.03% on SEED dataset and 97.31% on CAUEEG dataset for the classification tasks.","PeriodicalId":503691,"journal":{"name":"Machine Learning: Science and Technology","volume":"10 3","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-03-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140372510","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A Machine Learning Constitutive Model for Plasticity and Strain Hardening of Polycrystalline Metals Based on Data from Micromechanical Simulations","authors":"Ronak Shoghi, A. Hartmaier","doi":"10.1088/2632-2153/ad379e","DOIUrl":"https://doi.org/10.1088/2632-2153/ad379e","url":null,"abstract":"\u0000 Machine Learning (ML) methods have emerged as promising tools for generating constitutive models directly from mechanical data. Constitutive models are fundamental in describing and predicting the mechanical behavior of materials under arbitrary loading conditions. In recent approaches, the yield function, central to constitutive models, has been formulated in a data-oriented manner using ML. Many ML approaches have primarily focused on initial yielding, and the effect of strain hardening has not been widely considered. However, taking strain hardening into account is crucial for accurately describing the deformation behavior of polycrystalline metals. To address this problem, the present study introduces an ML-based yield function formulated as a Support Vector Classification (SVC) model, which encompasses strain hardening. This function was trained using a 12-dimensional feature vector that includes stress and plastic strain components resulting from crystal-plasticity finite element method (CPFEM) simulations on a three-dimensional representative volume element (RVE) with 343 grains with a random crystallographic texture. These simulations were carried out to mimic multi-axial mechanical testing of the polycrystal under proportional loading in 300 different directions, which were selected to ensure proper coverage of the full stress space. The training data were directly taken from the stress-strain results obtained for the 300 multi-axial load cases. It is shown that the ML yield function trained on these data describes not only the initial yield behavior but also the flow stresses in the plastic regime with a very high accuracy and robustness. The workflow introduced in this work to generate synthetic mechanical data based on realistic CPFEM simulations and to train an ML yield function, including strain hardening, will open new possibilities in microstructure-sensitive materials modeling and thus pave the way for obtaining digital material twins.","PeriodicalId":503691,"journal":{"name":"Machine Learning: Science and Technology","volume":" 561","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-03-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140382770","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Using Positional Tracking to Improve Abdominal Ultrasound Machine Learning Classification","authors":"Alistair Lawley, Rory Hampson, Kevin Worrall, Gordon Dobie","doi":"10.1088/2632-2153/ad379d","DOIUrl":"https://doi.org/10.1088/2632-2153/ad379d","url":null,"abstract":"\u0000 Diagnostic abdominal ultrasound screening and monitoring protocols are based around gathering a set of standard cross sectional images that ensure the coverage of relevant anatomical structures during the collection procedure. This allows clinicians to make diagnostic decisions with the best picture available from that modality. Currently, there is very little assistance provided to sonographers to ensure aderence to collection protocols, with previous studies suggesting that traditional image only machine learning classification can provide only limited assistance in supporting this task, for example it can be difficult to differentiate between multiple liver cross sections or those of the left and right kidney from image post collection. In this proof of concept, positional tracking information was added to the image input of a neural network to provide the additional context required to recognize six otherwise difficult to identify edge cases. In this paper optical and sensor based infrared tracking (IR) was used to track the position of an ultrasound probe during the collection of clinical cross sections on an abdominal phantom. Convolutional neural networks were then trained using both image-only and image with positional data, the classification accuracy results were then compared. The addition of positional information significantly improved average classification results from ~90% for image-only to 95% for optical IR position tracking and 93% for Sensor-based IR in common abdominal cross sections. While there is further work to be done, the addition of low-cost positional tracking to machine learning ultrasound classification will allow for significantly increased accuracy for identifying important diagnostic cross sections, with the potential to not only provide validation of adherence to protocol but also could provide navigation prompts to assist in user training and in ensuring aderence in capturing cross sections in future.","PeriodicalId":503691,"journal":{"name":"Machine Learning: Science and Technology","volume":"10 4","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-03-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140381550","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A method for quantifying the generalization capabilities of generative models for solving Ising models","authors":"Qunlong Ma, Zhi Ma, Ming Gao","doi":"10.1088/2632-2153/ad3710","DOIUrl":"https://doi.org/10.1088/2632-2153/ad3710","url":null,"abstract":"\u0000 For Ising models with complex energy landscapes, whether the ground state can be found by neural networks depends heavily on the Hamming distance between the training datasets and the ground state. Despite the fact that various recently proposed generative models have shown good performance in solving Ising models, there is no adequate discussion on how to quantify their generalization capabilities. Here we design a Hamming distance regularizer in the framework of a class of generative models, variational autoregressive networks (VAN), to quantify the generalization capabilities of various network architectures combined with VAN. The regularizer can control the size of the overlaps between the ground state and the training datasets generated by networks, which, together with the success rates of finding the ground state, form a quantitative metric to quantify their generalization capabilities. We conduct numerical experiments on several prototypical network architectures combined with VAN, including feed-forward neural networks, recurrent neural networks, and graph neural networks, to quantify their generalization capabilities when solving Ising models. Moreover, considering the fact that the quantification of the generalization capabilities of networks on small-scale problems can be used to predict their relative performance on large-scale problems, our method is of great significance for assisting in the Neural Architecture Search field of searching for the optimal network architectures when solving large-scale Ising models.","PeriodicalId":503691,"journal":{"name":"Machine Learning: Science and Technology","volume":" 4","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-03-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140212055","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Ensemble classifiers fed by functional connectivity during cognitive processing differentiate Parkinson’s disease even being under medication","authors":"E. Tülay","doi":"10.1088/2632-2153/ad370d","DOIUrl":"https://doi.org/10.1088/2632-2153/ad370d","url":null,"abstract":"\u0000 Objective: Brain-Computer Interface technologies, as a type of Human-Computer Interaction, provide a control ability on machines and intelligent systems via human brain functions without needing physical contact. Moreover, it has a considerable contribution to the detection of cognitive state changes, which gives a clue for neuropsychiatric diseases, including Parkinson’s disease (PD), in recent years. Although various studies implemented different machine learning models with several EEG features to detect PD and receive remarkable performances, there is a lack of knowledge on how brain connectivity during a cognitive task contributes to the differentiation of PD, even being under medication. Approach: To fill this gap, this study used three ensemble classifiers, which were fed by functional connectivity through cognitive response coherence (CRC) with varying selected features in different frequency bands upon application of the 3-Stimulation auditory oddball paradigm to differentiate PD medication ON and OFF and healthy controls (HC). Main results: The results revealed that the most remarkable performances were exhibited in slow frequency bands (delta and theta) in comparison to high frequency and wide range bands, especially in terms of target sounds. Moreover, in the delta band, target CRC distinguishes all groups from each other with accuracy rates of 80% for HC vs PD-OFF, 80% for HC vs PD-ON, and 81% for PD-ON vs PD-OFF. In the theta band, again target sounds were the most distinctive stimuli to classify HCxPD-OFF (80% accuracy), HCxPD-ON (80.5% accuracy) with quite good performances, and PD-ONxPD-OFF (76% accuracy) with acceptable performance. Besides, this study achieved a state-of-the-art performance with an accuracy of 87.5% in classifying PD-ON x PD-OFF via CRC of standard sounds in the delta band. Significance: Overall, the findings revealed that brain connectivity contributes to identifying PD and HC as well as the medication state of PD, especially in the slow frequency bands.","PeriodicalId":503691,"journal":{"name":"Machine Learning: Science and Technology","volume":" 82","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-03-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140213056","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Generation model meets Swin transformer for unsupervised low-dose CT reconstruction","authors":"Yu Li, Xueqin Sun, SuKai Wang, Yingwei Qin, Jinxiao Pan, Ping Chen","doi":"10.1088/2632-2153/ad370e","DOIUrl":"https://doi.org/10.1088/2632-2153/ad370e","url":null,"abstract":"\u0000 Computed Tomography (CT) has evolved into an indispensable tool for clinical diagnosis. Reducing radiation dose crucially minimizes adverse effects but may introduce noise and artifacts in reconstructed images, affecting diagnostic processes for physicians. Scholars have tackled deep learning training instability by exploring diffusion models. Given the scarcity of clinical data, we propose the Unsupervised Image Domain Score Generation model (UISG) for low-dose CT reconstruction. During training, normal-dose CT images are utilized as network inputs to train a score-based generative model that captures the prior distribution of CT images. In the iterative reconstruction, the initial CT image is obtained using a filtered back-projection algorithm. Subsequently, diffusion-based prior, high-frequency convolutional sparse coding prior, and data-consistency steps are employed to obtain the high-quality reconstructed image. Given the global characteristics of noise, the score network of the diffusion model utilizes a swin transformer structure to enhance the model's ability to capture long-range dependencies. Furthermore, convolutional sparse coding is applied exclusively to the high-frequency components of the image, to prevent over-smoothing or the loss of crucial anatomical details during the denoising process. Quantitative and qualitative results indicate that UISG outperforms competing methods in terms of denoising and generalization performance.","PeriodicalId":503691,"journal":{"name":"Machine Learning: Science and Technology","volume":" 11","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-03-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140217013","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Analysis and Benchmarking of Feature Reduction for Classification Under Computational Constraints","authors":"Omer Subasi, Sayan Ghosh, Joseph Manzano, Bruce Palmer, Andrés Marquez","doi":"10.1088/2632-2153/ad3726","DOIUrl":"https://doi.org/10.1088/2632-2153/ad3726","url":null,"abstract":"\u0000 Machine learning (ML) is most often expensive in terms of computational and memory costs due to training with large volumes of data. Current computational limitations of many computing systems motivate us to investigate practical approaches, such as feature selection and reduction, to reduce the time and memory costs while not sacrificing the accuracy of classification algorithms. In this work, we carefully review, analyze, and identify the feature reduction methods that have low costs/overheads in terms of time and memory. Then, we evaluate the identified reduction methods in terms of their impact on the accuracy, precision, time, and memory costs of traditional classification algorithms. Specifically, we focus on the least resource intensive feature reduction methods that are available in Scikit-Learn library. Since our goal is to identify the best performing low-cost reduction methods, we do not consider complex expensive reduction algorithms in this study. In our evaluation, we find that at quadratic-scale feature reduction, the classification algorithms achieve the best trade-off among competitive performance metrics. Results show that the overall training times are reduced 61%, the model sizes are reduced 6×, and accuracy scores increase 25% compared to the baselines on average with quadratic scale reduction.","PeriodicalId":503691,"journal":{"name":"Machine Learning: Science and Technology","volume":" 6","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-03-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140216742","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Graph Convolutional Multi-Mesh Autoencoder for Steady Transonic Aircraft Aerodynamics","authors":"David Massegur Sampietro, A. Da Ronch","doi":"10.1088/2632-2153/ad36ad","DOIUrl":"https://doi.org/10.1088/2632-2153/ad36ad","url":null,"abstract":"\u0000 Analysing the aerodynamic loads of an aircraft using computational fluid dynamics is a user's and computer-intensive task. An attractive alternative is to leverage on neural networks bypassing the need of solving the governing fluid equations at all flight conditions of interest. Neural networks have the ability to infer highly nonlinear predictions if a reference dataset is available. This work presents a geometric deep learning based multi-mesh autoencoder framework for steady-state transonic aerodynamics. The framework builds on graph neural networks which are designed for irregular and unstructured spatial discretisations, embedded in a multi-resolution algorithm for dimensionality reduction. The demonstration is for the NASA Common Research Model wing/body aircraft configuration. Thorough studies are presented discussing the model predictions in terms of vector fields, pressure and shear-stress coefficients, and scalar fields, total force and moment coefficients, for a range of nonlinear conditions involving shock waves and flow separation. We note that the cost of the model prediction is minimal, having used an existing available database.","PeriodicalId":503691,"journal":{"name":"Machine Learning: Science and Technology","volume":"198 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-03-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140222726","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}