Learning progressions divide the logical system of a subject into ordered and continuously developing levels that are suitable for the cognitive development level of students,which plays an important role in understan...Learning progressions divide the logical system of a subject into ordered and continuously developing levels that are suitable for the cognitive development level of students,which plays an important role in understanding students’learning process.This paper focuses on the theme of“kinetic energy”in high school physics as the research object.Firstly,the concept map was used to represent the relationship between knowledge,and then five core concepts were selected based on the opinions of high school teachers.Secondly,the test tools were compiled and tested based on the relevant test questions.Finally,the paper analyzed the results based on the Rasch model,clarified students’cognitive development level of“kinetic energy”and constructed the learning progressions of“kinetic energy”based on the logical order of subject knowledge.The research provides theoretical and methodological support for the study of other subjects and learning progressions,and provides a valuable reference for high school teachers to effectively carry out the instruction of“kinetic energy.”展开更多
Fine-scale structures can be observed in small field-of-view(FOV)auroral observations,but they are often overlooked because they appear only sporadically in all-sky observations.Such forms are of great interest becaus...Fine-scale structures can be observed in small field-of-view(FOV)auroral observations,but they are often overlooked because they appear only sporadically in all-sky observations.Such forms are of great interest because they may embody specific magnetosphere-ionosphere coupling processes,reveal localized energy deposition pathways,and provide new insights into cross-scale plasma dynamics and instabilities.However,their limited spatial extent,transient occurrence,and scarcity in wide-FOV observations make systematic investigation challenging.Traditional manual analysis struggles to capture these subtle structures within vast all-sky datasets,while automated detection faces severe data imbalance and morphological ambiguity.To address these challenges,we propose a synthetic-to-real progressive learning framework for cross-FOV retrieval of rare auroral forms.A Generative Adversarial Network(GAN)is employed to perform cross-FOV transformation between unpaired small-FOV images containing rare aurora forms and all-sky images(ASI)without such structures,thereby generating large numbers of synthetic ASI with rare auroral morphology.These synthetic samples are used to train an initial detection model,which subsequently undergoes iterative fine-tuning through feedback-guided learning:The model performs inference on new all-sky data,and the progressively accumulated real detections are incorporated into the training set.Experimental results demonstrate that the proposed method achieves over 92%detection accuracy on ASI,enabling high-precision retrieval of small-scale auroral structures across large-scale observations.This framework provides a scalable and effective approach to rediscovering rare auroral phenomena in continuous all-sky monitoring,offering new opportunities for exploring the fine-scale dynamics of the upper atmosphere.展开更多
In the past two decades, artificial intelligence （AI） algorithms have proved to be promising tools for solving several tough scientific problems, As a broad subfield of AI, machine learning is concerned with algorit...In the past two decades, artificial intelligence （AI） algorithms have proved to be promising tools for solving several tough scientific problems, As a broad subfield of AI, machine learning is concerned with algorithms and techniques that allow computers to ＂learn＂. The machine learning approach covers main domains such as data mining, difficult-to-program applications, and soft- ware applications. It is a collection of a variety of algorithms that can provide multivariate, nonlinear, nonparametric regression or classification. The remarkable simulation capabilities of the ma- chine learning-based methods have resulted in their extensive ap- plications in science and engineering. Recently, the machine learning techniques have found many applications in the geoscien- ces and remote sensing. More specifically, these techniques are proved to be practical for cases where the system＇s deterministic model is computationally expensive or there is no deterministic model to solve the problem （Lary, 2010）.展开更多
Computed tomography(CT)has enjoyed widespread applications,especially in the assistance of clinical diagnosis and treatment.However,fast CT imaging is not available for guiding adaptive precise radiotherapy in the cur...Computed tomography(CT)has enjoyed widespread applications,especially in the assistance of clinical diagnosis and treatment.However,fast CT imaging is not available for guiding adaptive precise radiotherapy in the current radiation treatment process because the conventional CT reconstruction requires numerous projections and rich computing resources.This paper mainly studies the challenging task of 3 D CT reconstruction from a single 2 D X-ray image of a particular patient,which enables fast CT imaging during radiotherapy.It is widely known that the transformation from a 2 D projection to a 3 D volumetric CT image is a highly nonlinear mapping problem.In this paper,we propose a progressive learning framework to facilitate 2 D-to-3 D mapping.The proposed network starts training from low resolution and then adds new layers to learn increasing high-resolution details as the training progresses.In addition,by bridging the distribution gap between an X-ray image and a CT image with a novel attention-based 2 D-to-3 D feature transform module and an adaptive instance normalization layer,our network obtains enhanced performance in recovering a 3 D CT volume from a single X-ray image.We demonstrate the effectiveness of our approach on a ten-phase 4 D CT dataset including 20 different patients created from a public medical database and show its outperformance over some baseline methods in image quality and structure preservation,achieving a PSNR value of 22.76±0.708 dB and FSIM value of 0.871±0.012 with the ground truth as a reference.This method may promote the application of CT imaging in adaptive radiotherapy and provide image guidance for interventional surgery.展开更多
Federated learning is an effective distributed learning framework that protects privacy and allows multiple edge devices to work together to train models jointly without exchanging data.However,edge devices usually ha...Federated learning is an effective distributed learning framework that protects privacy and allows multiple edge devices to work together to train models jointly without exchanging data.However,edge devices usually have limited com-puting capabilities,and limited network bandwidth is often a major bottleneck.In order to reduce communication and computing costs,we introduced a horizon-tal pruning mechanism,combined federated learning and progressive learning,and proposed a progressive federated learning scheme based on model pruning.It gradually trains from simple models to more complex ones and trims the uploaded models horizontally.Our approach effectively reduces computational and bidirec-tional communication costs while maintaining model performance.Several image classification experiments on different models have been conducted by us,and the experimental results demonstrate that our approach can effectively save approxi-mately 10%of the computational cost and 48%of the communication cost when compared to FedAvg.展开更多
基金Jilin Province Education Science“14th Five Year Plan”2021 Annual Project“Research on Middle School Physics Teaching Based on STEM Education Concept”(GH21032).
文摘Learning progressions divide the logical system of a subject into ordered and continuously developing levels that are suitable for the cognitive development level of students,which plays an important role in understanding students’learning process.This paper focuses on the theme of“kinetic energy”in high school physics as the research object.Firstly,the concept map was used to represent the relationship between knowledge,and then five core concepts were selected based on the opinions of high school teachers.Secondly,the test tools were compiled and tested based on the relevant test questions.Finally,the paper analyzed the results based on the Rasch model,clarified students’cognitive development level of“kinetic energy”and constructed the learning progressions of“kinetic energy”based on the logical order of subject knowledge.The research provides theoretical and methodological support for the study of other subjects and learning progressions,and provides a valuable reference for high school teachers to effectively carry out the instruction of“kinetic energy.”
基金supported by the National Natural Science Foundation of China(Grant no.41874173).
文摘Fine-scale structures can be observed in small field-of-view(FOV)auroral observations,but they are often overlooked because they appear only sporadically in all-sky observations.Such forms are of great interest because they may embody specific magnetosphere-ionosphere coupling processes,reveal localized energy deposition pathways,and provide new insights into cross-scale plasma dynamics and instabilities.However,their limited spatial extent,transient occurrence,and scarcity in wide-FOV observations make systematic investigation challenging.Traditional manual analysis struggles to capture these subtle structures within vast all-sky datasets,while automated detection faces severe data imbalance and morphological ambiguity.To address these challenges,we propose a synthetic-to-real progressive learning framework for cross-FOV retrieval of rare auroral forms.A Generative Adversarial Network(GAN)is employed to perform cross-FOV transformation between unpaired small-FOV images containing rare aurora forms and all-sky images(ASI)without such structures,thereby generating large numbers of synthetic ASI with rare auroral morphology.These synthetic samples are used to train an initial detection model,which subsequently undergoes iterative fine-tuning through feedback-guided learning:The model performs inference on new all-sky data,and the progressively accumulated real detections are incorporated into the training set.Experimental results demonstrate that the proposed method achieves over 92%detection accuracy on ASI,enabling high-precision retrieval of small-scale auroral structures across large-scale observations.This framework provides a scalable and effective approach to rediscovering rare auroral phenomena in continuous all-sky monitoring,offering new opportunities for exploring the fine-scale dynamics of the upper atmosphere.
文摘In the past two decades, artificial intelligence （AI） algorithms have proved to be promising tools for solving several tough scientific problems, As a broad subfield of AI, machine learning is concerned with algorithms and techniques that allow computers to ＂learn＂. The machine learning approach covers main domains such as data mining, difficult-to-program applications, and soft- ware applications. It is a collection of a variety of algorithms that can provide multivariate, nonlinear, nonparametric regression or classification. The remarkable simulation capabilities of the ma- chine learning-based methods have resulted in their extensive ap- plications in science and engineering. Recently, the machine learning techniques have found many applications in the geoscien- ces and remote sensing. More specifically, these techniques are proved to be practical for cases where the system＇s deterministic model is computationally expensive or there is no deterministic model to solve the problem （Lary, 2010）.
文摘Computed tomography(CT)has enjoyed widespread applications,especially in the assistance of clinical diagnosis and treatment.However,fast CT imaging is not available for guiding adaptive precise radiotherapy in the current radiation treatment process because the conventional CT reconstruction requires numerous projections and rich computing resources.This paper mainly studies the challenging task of 3 D CT reconstruction from a single 2 D X-ray image of a particular patient,which enables fast CT imaging during radiotherapy.It is widely known that the transformation from a 2 D projection to a 3 D volumetric CT image is a highly nonlinear mapping problem.In this paper,we propose a progressive learning framework to facilitate 2 D-to-3 D mapping.The proposed network starts training from low resolution and then adds new layers to learn increasing high-resolution details as the training progresses.In addition,by bridging the distribution gap between an X-ray image and a CT image with a novel attention-based 2 D-to-3 D feature transform module and an adaptive instance normalization layer,our network obtains enhanced performance in recovering a 3 D CT volume from a single X-ray image.We demonstrate the effectiveness of our approach on a ten-phase 4 D CT dataset including 20 different patients created from a public medical database and show its outperformance over some baseline methods in image quality and structure preservation,achieving a PSNR value of 22.76±0.708 dB and FSIM value of 0.871±0.012 with the ground truth as a reference.This method may promote the application of CT imaging in adaptive radiotherapy and provide image guidance for interventional surgery.
文摘Federated learning is an effective distributed learning framework that protects privacy and allows multiple edge devices to work together to train models jointly without exchanging data.However,edge devices usually have limited com-puting capabilities,and limited network bandwidth is often a major bottleneck.In order to reduce communication and computing costs,we introduced a horizon-tal pruning mechanism,combined federated learning and progressive learning,and proposed a progressive federated learning scheme based on model pruning.It gradually trains from simple models to more complex ones and trims the uploaded models horizontally.Our approach effectively reduces computational and bidirec-tional communication costs while maintaining model performance.Several image classification experiments on different models have been conducted by us,and the experimental results demonstrate that our approach can effectively save approxi-mately 10%of the computational cost and 48%of the communication cost when compared to FedAvg.
