Ensuring the consistent mechanical performance of three-dimensional(3D)-printed continuous fiber-reinforced composites is a significant challenge in additive manufacturing.The current reliance on manual monitoring exa...Ensuring the consistent mechanical performance of three-dimensional(3D)-printed continuous fiber-reinforced composites is a significant challenge in additive manufacturing.The current reliance on manual monitoring exacerbates this challenge by rendering the process vulnerable to environmental changes and unexpected factors,resulting in defects and inconsistent product quality,particularly in unmanned long-term operations or printing in extreme environments.To address these issues,we developed a process monitoring and closed-loop feedback control strategy for the 3D printing process.Real-time printing image data were captured and analyzed using a well-trained neural network model,and a real-time control module-enabled closed-loop feedback control of the flow rate was developed.The neural network model,which was based on image processing and artificial intelligence,enabled the recognition of flow rate values with an accuracy of 94.70%.The experimental results showed significant improvements in both the surface performance and mechanical properties of printed composites,with three to six times improvement in tensile strength and elastic modulus,demonstrating the effectiveness of the strategy.This study provides a generalized process monitoring and feedback control method for the 3D printing of continuous fiber-reinforced composites,and offers a potential solution for remote online monitoring and closed-loop adjustment in unmanned or extreme space environments.展开更多
In natural language processing(NLP),managing multiple downstream tasks through fine-tuning pre-trained models often requires maintaining separate task-specific models,leading to practical inefficiencies.To address thi...In natural language processing(NLP),managing multiple downstream tasks through fine-tuning pre-trained models often requires maintaining separate task-specific models,leading to practical inefficiencies.To address this challenge,we introduce AdaptForever,a novel approach that enables continuous mastery of NLP tasks through the integration of elastic and mutual learning strategies with a stochastic expert mechanism.Our method freezes the pre-trained model weights while incorporating adapters enhanced with mutual learning capabilities,facilitating effective knowledge transfer from previous tasks to new ones.By combining Elastic Weight Consolidation(EWC)for knowledge preservation with specialized regularization terms,AdaptForever successfully maintains performance on earlier tasks while acquiring new capabilities.Experimental results demonstrate that AdaptForever achieves superior performance across a continuous sequence of NLP tasks compared to existing parameter-efficient methods,while effectively preventing catastrophic forgetting and enabling positive knowledge transfer between tasks.展开更多
Recently,machine learning has become a powerful tool for predicting nuclear charge radius RC,providing novel insights into complex physical phenomena.This study employs a continuous Bayesian probability(CBP)estimator ...Recently,machine learning has become a powerful tool for predicting nuclear charge radius RC,providing novel insights into complex physical phenomena.This study employs a continuous Bayesian probability(CBP)estimator and Bayesian model averaging(BMA)to optimize the predictions of RCfrom sophisticated theoretical models.The CBP estimator treats the residual between the theoretical and experimental values of RCas a continuous variable and derives its posterior probability density function(PDF)from Bayesian theory.The BMA method assigns weights to models based on their predictive performance for benchmark nuclei,thereby accounting for the unique strengths of each model.In global optimization,the CBP estimator improved the predictive accuracy of the three theoretical models by approximately 60%.The extrapolation analyses consistently achieved an improvement rate of approximately 45%,demonstrating the robustness of the CBP estimator.Furthermore,the combination of the CBP and BMA methods reduces the standard deviation to below 0.02 fm,effectively reproducing the pronounced shell effects on RCof the Ca and Sr isotope chains.The studies in this paper propose an efficient method to accurately describe RCof unknown nuclei,with potential applications in research on other nuclear properties.展开更多
Although modulation classification based on deep neural network can achieve high Modulation Classification(MC)accuracies,catastrophic forgetting will occur when the neural network model continues to learn new tasks.In...Although modulation classification based on deep neural network can achieve high Modulation Classification(MC)accuracies,catastrophic forgetting will occur when the neural network model continues to learn new tasks.In this paper,we simulate the dynamic wireless communication environment and focus on breaking the learning paradigm of isolated automatic MC.We innovate a research algorithm for continuous automatic MC.Firstly,a memory for storing representative old task modulation signals is built,which is employed to limit the gradient update direction of new tasks in the continuous learning stage to ensure that the loss of old tasks is also in a downward trend.Secondly,in order to better simulate the dynamic wireless communication environment,we employ the mini-batch gradient algorithm which is more suitable for continuous learning.Finally,the signal in the memory can be replayed to further strengthen the characteristics of the old task signal in the model.Simulation results verify the effectiveness of the method.展开更多
1|OVERVIEW.Machine learning(ML)has been increasingly used for tackling various diagnostic,therapeutic,and prognostic tasks owing to its capability to learn and reason without explicit programming[1].Most developed ML ...1|OVERVIEW.Machine learning(ML)has been increasingly used for tackling various diagnostic,therapeutic,and prognostic tasks owing to its capability to learn and reason without explicit programming[1].Most developed ML models have had their accuracy proven through internal validation using retrospective data.However,external validation using retrospective data,continual monitoring using prospective data,and randomized controlled trials(RCTs)using prospective data are important for the translation of ML models into real-world clinical practice[2].展开更多
An iterative learning model predictive control （ILMPC） technique is applied to a class of continuous/batch processes. Such processes are characterized by the operations of batch processes generating periodic strong ...An iterative learning model predictive control （ILMPC） technique is applied to a class of continuous/batch processes. Such processes are characterized by the operations of batch processes generating periodic strong disturbances to the continuous processes and traditional regulatory controllers are unable to eliminate these periodic disturbances. ILMPC integrates the feature of iterative learning control （ILC） handling repetitive signal and the flexibility of model predictive control （MPC）. By on-line monitoring the operation status of batch processes, an event-driven iterative learning algorithm for batch repetitive disturbances is initiated and the soft constraints are adjusted timely as the feasible region is away from the desired operating zone. The results of an industrial application show that the proposed ILMPC method is effective for a class of continuous/batch processes.展开更多
Climate change poses significant challenges to agricultural management,particularly in adapting to extreme weather conditions that impact agricultural production.Existing works with traditional Reinforcement Learning(...Climate change poses significant challenges to agricultural management,particularly in adapting to extreme weather conditions that impact agricultural production.Existing works with traditional Reinforcement Learning(RL)methods often falter under such extreme conditions.To address this challenge,our study introduces a novel approach by integrating Continual Learning(CL)with RL to form Continual Reinforcement Learning(CRL),enhancing the adaptability of agricultural management strategies.Leveraging the Gym-DSSAT simulation environment,our research enables RL agents to learn optimal fertilization strategies based on variable weather conditions.By incorporating CL algorithms,such as Elastic Weight Consolidation(EWC),with established RL techniques like Deep Q-Networks(DQN),we developed a framework in which agents can learn and retain knowledge across diverse weather scenarios.The CRL approach was tested under climate variability to assess the robustness and adaptability of the induced policies,particularly under extreme weather events like severe droughts.Our results showed that continually learned policies exhibited superior adaptability and performance compared to optimal policies learned through the conventional RL methods,especially in challenging conditions of reduced rainfall and increased temperatures.This pioneering work,which combines CL with RL to generate adaptive policies for agricultural management,is expected to make significant advancements in precision agriculture in the era of climate change.展开更多
Reinforcement Learning is a commonly used technique for learning tasks in robotics, however, traditional algorithms are unable to handle large amounts of data coming from the robot’s sensors, require long training ti...Reinforcement Learning is a commonly used technique for learning tasks in robotics, however, traditional algorithms are unable to handle large amounts of data coming from the robot’s sensors, require long training times, and use dis-crete actions. This work introduces TS-RRLCA, a two stage method to tackle these problems. In the first stage, low-level data coming from the robot’s sensors is transformed into a more natural, relational representation based on rooms, walls, corners, doors and obstacles, significantly reducing the state space. We use this representation along with Behavioural Cloning, i.e., traces provided by the user;to learn, in few iterations, a relational control policy with discrete actions which can be re-used in different environments. In the second stage, we use Locally Weighted Regression to transform the initial policy into a continuous actions policy. We tested our approach in simulation and with a real service robot on different environments for different navigation and following tasks. Results show how the policies can be used on different domains and perform smoother, faster and shorter paths than the original discrete actions policies.展开更多
The analytic continuation serves as a crucial bridge between quantum Monte Carlo calculations in imaginary-time formalism,specifically the Green's functions,and physical measurements(the spectral functions)in real...The analytic continuation serves as a crucial bridge between quantum Monte Carlo calculations in imaginary-time formalism,specifically the Green's functions,and physical measurements(the spectral functions)in real time.Various approaches have been developed to enhance the accuracy of analytic continuation,including the Padéapproximation,the maximum entropy method,and stochastic analytic continuation.In this study,we employ different deep learning techniques to investigate the analytic continuation for the quantum impurity model.A significant challenge in this context is that the sharp Abrikosov-Suhl resonance peak may be either underestimated or overestimated.We fit both the imaginary-time Green's function and the spectral function using Chebyshev polynomials in logarithmic coordinates.We utilize Full-Connected Networks(FCN),Convolutional Neural Networks(CNNs),and Residual Networks(ResNet)to address this issue.Our findings indicate that introducing noise during the training phase significantly improves the accuracy of the learning process.The typical absolute error achieved is less than 10-4.These investigations pave the way for machine learning to optimize the analytic continuation problem in many-body systems,thereby reducing the need for prior expertise in physics.展开更多
Graphs have been widely used in fields ranging from chemical informatics to social network analysis.Graph-related problems become increasingly significant,with subgraph matching standing out as one of the most challen...Graphs have been widely used in fields ranging from chemical informatics to social network analysis.Graph-related problems become increasingly significant,with subgraph matching standing out as one of the most challenging tasks.The goal of subgraph matching is to find all subgraphs in the data graph that are isomorphic to the query graph.Traditional methods mostly rely on search strategies with high computational complexity and are hard to apply to large-scale real datasets.With the advent of graph neural networks(GNNs),researchers have turned to GNNs to address subgraph matching problems.However,the multi-attributed features on nodes and edges are overlooked during the learning of graphs,which causes inaccurate results in real-world scenarios.To tackle this problem,we propose a novel model called subgraph matching on multi-attributed graph network(SGMAN).SGMAN first utilizes improved line graphs to capture node and edge features.Then,SGMAN integrates GNN and contrastive learning(CL)to derive graph representation embeddings and calculate the matching matrix to represent the matching results.We conduct experiments on public datasets,and the results affirm the superior performance of our model.展开更多
Continual learning(CL)studies the problem of learning to accumulate knowledge over time from a stream of data.A crucial challenge is that neural networks suffer from performance degradation on previously seen data,kno...Continual learning(CL)studies the problem of learning to accumulate knowledge over time from a stream of data.A crucial challenge is that neural networks suffer from performance degradation on previously seen data,known as catastrophic forgetting,due to allowing parameter sharing.In this work,we consider a more practical online class-incremental CL setting,where the model learns new samples in an online manner and may continuously experience new classes.Moreover,prior knowledge is unavailable during training and evaluation.Existing works usually explore sample usages from a single dimension,which ignores a lot of valuable supervisory information.To better tackle the setting,we propose a novel replay-based CL method,which leverages multi-level representations produced by the intermediate process of training samples for replay and strengthens supervision to consolidate previous knowledge.Specifically,besides the previous raw samples,we store the corresponding logits and features in the memory.Furthermore,to imitate the prediction of the past model,we construct extra constraints by leveraging multi-level information stored in the memory.With the same number of samples for replay,our method can use more past knowledge to prevent interference.We conduct extensive evaluations on several popular CL datasets,and experiments show that our method consistently outperforms state-of-the-art methods with various sizes of episodic memory.We further provide a detailed analysis of these results and demonstrate that our method is more viable in practical scenarios.展开更多
Continuous cooling transformation diagrams in synthetic weld heat-affected zone(SH-CCT diagrams)show the phase transition temperature and hardness at different cooling rates,which is an important basis for formulating...Continuous cooling transformation diagrams in synthetic weld heat-affected zone(SH-CCT diagrams)show the phase transition temperature and hardness at different cooling rates,which is an important basis for formulating the welding process or predicting the performance of welding heat-affected zone.However,the experimental determination of SH-CCT diagrams is a time-consuming and costly process,which does not conform to the development trend of new materials.In addition,the prediction of SHCCT diagrams using metallurgical models remains a challenge due to the complexity of alloying elements and welding processes.So,in this study,a hybrid machine learning model consisting of multilayer perceptron classifier,k-Nearest Neighbors and random forest is established to predict the phase transformation temperature and hardness of low alloy steel using chemical composition and cooling rate.Then the SH-CCT diagrams of 6 kinds of steels are calculated by the hybrid machine learning model.The results show that the accuracy of the classification model is up to 100%,the predicted values of the regression models are in good agreement with the experimental results,with high correlation coefficient and low error value.Moreover,the mathematical expressions of hardness in welding heat-affected zone of low alloy steel are calculated by symbolic regression,which can quantitatively express the relationship between alloy composition,cooling time and hardness.This study demonstrates the great potential of the material informatics in the field of welding technology.展开更多
The predominant method for smart phone accessing is confined to methods directing the authentication by means of Point-of-Entry that heavily depend on physiological biometrics like,fingerprint or face.Implicit continuou...The predominant method for smart phone accessing is confined to methods directing the authentication by means of Point-of-Entry that heavily depend on physiological biometrics like,fingerprint or face.Implicit continuous authentication initiating to be loftier to conventional authentication mechanisms by continuously confirming users’identities on continuing basis and mark the instant at which an illegitimate hacker grasps dominance of the session.However,divergent issues remain unaddressed.This research aims to investigate the power of Deep Reinforcement Learning technique to implicit continuous authentication for mobile devices using a method called,Gaussian Weighted Cauchy Kriging-based Continuous Czekanowski’s(GWCK-CC).First,a Gaussian Weighted Non-local Mean Filter Preprocessing model is applied for reducing the noise pre-sent in the raw input face images.Cauchy Kriging Regression function is employed to reduce the dimensionality.Finally,Continuous Czekanowski’s Clas-sification is utilized for proficient classification between the genuine user and attacker.By this way,the proposed GWCK-CC method achieves accurate authen-tication with minimum error rate and time.Experimental assessment of the pro-posed GWCK-CC method and existing methods are carried out with different factors by using UMDAA-02 Face Dataset.The results confirm that the proposed GWCK-CC method enhances authentication accuracy,by 9%,reduces the authen-tication time,and error rate by 44%,and 43%as compared to the existing methods.展开更多
This study proposed a measurement platform for continuous blood pressure estimation based on dual photoplethysmography(PPG)sensors and a deep learning(DL)that can be used for continuous and rapid measurement of blood ...This study proposed a measurement platform for continuous blood pressure estimation based on dual photoplethysmography(PPG)sensors and a deep learning(DL)that can be used for continuous and rapid measurement of blood pressure and analysis of cardiovascular-related indicators.The proposed platform measured the signal changes in PPG and converted them into physiological indicators,such as pulse transit time(PTT),pulse wave velocity(PWV),perfusion index(PI)and heart rate(HR);these indicators were then fed into the DL to calculate blood pressure.The hardware of the experiment comprised 2 PPG components(i.e.,Raspberry Pi 3 Model B and analog-todigital converter[MCP3008]),which were connected using a serial peripheral interface.The DL algorithm converted the stable dual PPG signals acquired from the strictly standardized experimental process into various physiological indicators as input parameters and finally obtained the systolic blood pressure(SBP),diastolic blood pressure(DBP)and mean arterial pressure(MAP).To increase the robustness of the DL model,this study input data of 100 Asian participants into the training database,including those with and without cardiovascular disease,each with a proportion of approximately 50%.The experimental results revealed that the mean absolute error and standard deviation of SBP was 0.17±0.46 mmHg.The mean absolute error and standard deviation of DBP was 0.27±0.52 mmHg.The mean absolute error and standard deviation of MAP was 0.16±0.40 mmHg.展开更多
The overall research in Reinforcement Learning (RL) concentrates on discrete sets of actions, but for certain real-world problems it is important to have methods which are able to find good strategies using actions dr...The overall research in Reinforcement Learning (RL) concentrates on discrete sets of actions, but for certain real-world problems it is important to have methods which are able to find good strategies using actions drawn from continuous sets. This paper describes a simple control task called direction finder and its known optimal solution for both discrete and continuous actions. It allows for comparison of RL solution methods based on their value functions. In order to solve the control task for continuous actions, a simple idea for generalising them by means of feature vectors is presented. The resulting algorithm is applied using different choices of feature calculations. For comparing their performance a simple measure is展开更多
Cooperative learning emerging as the leading new approach to classroom instruction abroad over the past decades has been studied by many researchers from all aspects.This paper mainly focuses on the basics of cooperat...Cooperative learning emerging as the leading new approach to classroom instruction abroad over the past decades has been studied by many researchers from all aspects.This paper mainly focuses on the basics of cooperative learning and tries to answer the question that if the use of cooperative learning produce higher achievement than the traditional methods in college English reading class through experimental study.The analysis contributes to better college English teaching and learning.A conclusion is drawn that cooperative learning is very effective in improving college students reading ability.展开更多
基金supported by National Key Research and Development Program of China(Grant No.2023YFB4604100)National Key Research and Development Program of China(Grant No.2022YFB3806104)+4 种基金Key Research and Development Program in Shaanxi Province(Grant No.2021LLRH-08-17)Young Elite Scientists Sponsorship Program by CAST(No.2023QNRC001)K C Wong Education Foundation of ChinaYouth Innovation Team of Shaanxi Universities of ChinaKey Research and Development Program of Shaanxi Province(Grant 2021LLRH-08-3.1).
文摘Ensuring the consistent mechanical performance of three-dimensional(3D)-printed continuous fiber-reinforced composites is a significant challenge in additive manufacturing.The current reliance on manual monitoring exacerbates this challenge by rendering the process vulnerable to environmental changes and unexpected factors,resulting in defects and inconsistent product quality,particularly in unmanned long-term operations or printing in extreme environments.To address these issues,we developed a process monitoring and closed-loop feedback control strategy for the 3D printing process.Real-time printing image data were captured and analyzed using a well-trained neural network model,and a real-time control module-enabled closed-loop feedback control of the flow rate was developed.The neural network model,which was based on image processing and artificial intelligence,enabled the recognition of flow rate values with an accuracy of 94.70%.The experimental results showed significant improvements in both the surface performance and mechanical properties of printed composites,with three to six times improvement in tensile strength and elastic modulus,demonstrating the effectiveness of the strategy.This study provides a generalized process monitoring and feedback control method for the 3D printing of continuous fiber-reinforced composites,and offers a potential solution for remote online monitoring and closed-loop adjustment in unmanned or extreme space environments.
基金supported by the National Key R&D Program of China(No.2023YFB3308601)Sichuan Science and Technology Program(2024NSFJQ0035,2024NSFSC0004)the Talents by Sichuan provincial Party Committee Organization Department.
文摘In natural language processing(NLP),managing multiple downstream tasks through fine-tuning pre-trained models often requires maintaining separate task-specific models,leading to practical inefficiencies.To address this challenge,we introduce AdaptForever,a novel approach that enables continuous mastery of NLP tasks through the integration of elastic and mutual learning strategies with a stochastic expert mechanism.Our method freezes the pre-trained model weights while incorporating adapters enhanced with mutual learning capabilities,facilitating effective knowledge transfer from previous tasks to new ones.By combining Elastic Weight Consolidation(EWC)for knowledge preservation with specialized regularization terms,AdaptForever successfully maintains performance on earlier tasks while acquiring new capabilities.Experimental results demonstrate that AdaptForever achieves superior performance across a continuous sequence of NLP tasks compared to existing parameter-efficient methods,while effectively preventing catastrophic forgetting and enabling positive knowledge transfer between tasks.
基金supported by the National Natural Science Foundation of China(Nos.12475135,12035011,and 12475119)the Shandong Provincial Natural Science Foundation,China(No.ZR2020MA096)the Fundamental Research Funds for the Central Universities(No.22CX03017A)。
文摘Recently,machine learning has become a powerful tool for predicting nuclear charge radius RC,providing novel insights into complex physical phenomena.This study employs a continuous Bayesian probability(CBP)estimator and Bayesian model averaging(BMA)to optimize the predictions of RCfrom sophisticated theoretical models.The CBP estimator treats the residual between the theoretical and experimental values of RCas a continuous variable and derives its posterior probability density function(PDF)from Bayesian theory.The BMA method assigns weights to models based on their predictive performance for benchmark nuclei,thereby accounting for the unique strengths of each model.In global optimization,the CBP estimator improved the predictive accuracy of the three theoretical models by approximately 60%.The extrapolation analyses consistently achieved an improvement rate of approximately 45%,demonstrating the robustness of the CBP estimator.Furthermore,the combination of the CBP and BMA methods reduces the standard deviation to below 0.02 fm,effectively reproducing the pronounced shell effects on RCof the Ca and Sr isotope chains.The studies in this paper propose an efficient method to accurately describe RCof unknown nuclei,with potential applications in research on other nuclear properties.
文摘Although modulation classification based on deep neural network can achieve high Modulation Classification(MC)accuracies,catastrophic forgetting will occur when the neural network model continues to learn new tasks.In this paper,we simulate the dynamic wireless communication environment and focus on breaking the learning paradigm of isolated automatic MC.We innovate a research algorithm for continuous automatic MC.Firstly,a memory for storing representative old task modulation signals is built,which is employed to limit the gradient update direction of new tasks in the continuous learning stage to ensure that the loss of old tasks is also in a downward trend.Secondly,in order to better simulate the dynamic wireless communication environment,we employ the mini-batch gradient algorithm which is more suitable for continuous learning.Finally,the signal in the memory can be replayed to further strengthen the characteristics of the old task signal in the model.Simulation results verify the effectiveness of the method.
文摘1|OVERVIEW.Machine learning(ML)has been increasingly used for tackling various diagnostic,therapeutic,and prognostic tasks owing to its capability to learn and reason without explicit programming[1].Most developed ML models have had their accuracy proven through internal validation using retrospective data.However,external validation using retrospective data,continual monitoring using prospective data,and randomized controlled trials(RCTs)using prospective data are important for the translation of ML models into real-world clinical practice[2].
基金Supported by the National Creative Research Groups Science Foundation of China （60721062） and the National High Technology Research and Development Program of China （2007AA04Z162）.
文摘An iterative learning model predictive control （ILMPC） technique is applied to a class of continuous/batch processes. Such processes are characterized by the operations of batch processes generating periodic strong disturbances to the continuous processes and traditional regulatory controllers are unable to eliminate these periodic disturbances. ILMPC integrates the feature of iterative learning control （ILC） handling repetitive signal and the flexibility of model predictive control （MPC）. By on-line monitoring the operation status of batch processes, an event-driven iterative learning algorithm for batch repetitive disturbances is initiated and the soft constraints are adjusted timely as the feasible region is away from the desired operating zone. The results of an industrial application show that the proposed ILMPC method is effective for a class of continuous/batch processes.
基金support from the University of Iowa OVPR Interdisciplinary Scholars Program and the US Department of Education(ED#P116S210005)for this study.Kishlay Jha’s work is supported in part by the US National Institute of Health(NIH)and National Science Foundation(NSF)under grants R01LM014012-01A1 and ITE-2333740.
文摘Climate change poses significant challenges to agricultural management,particularly in adapting to extreme weather conditions that impact agricultural production.Existing works with traditional Reinforcement Learning(RL)methods often falter under such extreme conditions.To address this challenge,our study introduces a novel approach by integrating Continual Learning(CL)with RL to form Continual Reinforcement Learning(CRL),enhancing the adaptability of agricultural management strategies.Leveraging the Gym-DSSAT simulation environment,our research enables RL agents to learn optimal fertilization strategies based on variable weather conditions.By incorporating CL algorithms,such as Elastic Weight Consolidation(EWC),with established RL techniques like Deep Q-Networks(DQN),we developed a framework in which agents can learn and retain knowledge across diverse weather scenarios.The CRL approach was tested under climate variability to assess the robustness and adaptability of the induced policies,particularly under extreme weather events like severe droughts.Our results showed that continually learned policies exhibited superior adaptability and performance compared to optimal policies learned through the conventional RL methods,especially in challenging conditions of reduced rainfall and increased temperatures.This pioneering work,which combines CL with RL to generate adaptive policies for agricultural management,is expected to make significant advancements in precision agriculture in the era of climate change.
文摘Reinforcement Learning is a commonly used technique for learning tasks in robotics, however, traditional algorithms are unable to handle large amounts of data coming from the robot’s sensors, require long training times, and use dis-crete actions. This work introduces TS-RRLCA, a two stage method to tackle these problems. In the first stage, low-level data coming from the robot’s sensors is transformed into a more natural, relational representation based on rooms, walls, corners, doors and obstacles, significantly reducing the state space. We use this representation along with Behavioural Cloning, i.e., traces provided by the user;to learn, in few iterations, a relational control policy with discrete actions which can be re-used in different environments. In the second stage, we use Locally Weighted Regression to transform the initial policy into a continuous actions policy. We tested our approach in simulation and with a real service robot on different environments for different navigation and following tasks. Results show how the policies can be used on different domains and perform smoother, faster and shorter paths than the original discrete actions policies.
基金Sponsored by National Natural Science Foundation of China(Grant No.12174101)Fundamental Research Funds for the Central Universities(Grant No.2022MS051).
文摘The analytic continuation serves as a crucial bridge between quantum Monte Carlo calculations in imaginary-time formalism,specifically the Green's functions,and physical measurements(the spectral functions)in real time.Various approaches have been developed to enhance the accuracy of analytic continuation,including the Padéapproximation,the maximum entropy method,and stochastic analytic continuation.In this study,we employ different deep learning techniques to investigate the analytic continuation for the quantum impurity model.A significant challenge in this context is that the sharp Abrikosov-Suhl resonance peak may be either underestimated or overestimated.We fit both the imaginary-time Green's function and the spectral function using Chebyshev polynomials in logarithmic coordinates.We utilize Full-Connected Networks(FCN),Convolutional Neural Networks(CNNs),and Residual Networks(ResNet)to address this issue.Our findings indicate that introducing noise during the training phase significantly improves the accuracy of the learning process.The typical absolute error achieved is less than 10-4.These investigations pave the way for machine learning to optimize the analytic continuation problem in many-body systems,thereby reducing the need for prior expertise in physics.
文摘Graphs have been widely used in fields ranging from chemical informatics to social network analysis.Graph-related problems become increasingly significant,with subgraph matching standing out as one of the most challenging tasks.The goal of subgraph matching is to find all subgraphs in the data graph that are isomorphic to the query graph.Traditional methods mostly rely on search strategies with high computational complexity and are hard to apply to large-scale real datasets.With the advent of graph neural networks(GNNs),researchers have turned to GNNs to address subgraph matching problems.However,the multi-attributed features on nodes and edges are overlooked during the learning of graphs,which causes inaccurate results in real-world scenarios.To tackle this problem,we propose a novel model called subgraph matching on multi-attributed graph network(SGMAN).SGMAN first utilizes improved line graphs to capture node and edge features.Then,SGMAN integrates GNN and contrastive learning(CL)to derive graph representation embeddings and calculate the matching matrix to represent the matching results.We conduct experiments on public datasets,and the results affirm the superior performance of our model.
基金supported in part by the National Natura Science Foundation of China(U2013602,61876181,51521003)the Nationa Key R&D Program of China(2020YFB13134)+2 种基金Shenzhen Science and Technology Research and Development Foundation(JCYJ20190813171009236)Beijing Nova Program of Science and Technology(Z191100001119043)the Youth Innovation Promotion Association,Chinese Academy of Sciences。
文摘Continual learning(CL)studies the problem of learning to accumulate knowledge over time from a stream of data.A crucial challenge is that neural networks suffer from performance degradation on previously seen data,known as catastrophic forgetting,due to allowing parameter sharing.In this work,we consider a more practical online class-incremental CL setting,where the model learns new samples in an online manner and may continuously experience new classes.Moreover,prior knowledge is unavailable during training and evaluation.Existing works usually explore sample usages from a single dimension,which ignores a lot of valuable supervisory information.To better tackle the setting,we propose a novel replay-based CL method,which leverages multi-level representations produced by the intermediate process of training samples for replay and strengthens supervision to consolidate previous knowledge.Specifically,besides the previous raw samples,we store the corresponding logits and features in the memory.Furthermore,to imitate the prediction of the past model,we construct extra constraints by leveraging multi-level information stored in the memory.With the same number of samples for replay,our method can use more past knowledge to prevent interference.We conduct extensive evaluations on several popular CL datasets,and experiments show that our method consistently outperforms state-of-the-art methods with various sizes of episodic memory.We further provide a detailed analysis of these results and demonstrate that our method is more viable in practical scenarios.
基金financial support from the National Key Research and Development Program of China[No.2016YFB0700501]the National Natural Science Foundation of China(No.51571020)。
文摘Continuous cooling transformation diagrams in synthetic weld heat-affected zone(SH-CCT diagrams)show the phase transition temperature and hardness at different cooling rates,which is an important basis for formulating the welding process or predicting the performance of welding heat-affected zone.However,the experimental determination of SH-CCT diagrams is a time-consuming and costly process,which does not conform to the development trend of new materials.In addition,the prediction of SHCCT diagrams using metallurgical models remains a challenge due to the complexity of alloying elements and welding processes.So,in this study,a hybrid machine learning model consisting of multilayer perceptron classifier,k-Nearest Neighbors and random forest is established to predict the phase transformation temperature and hardness of low alloy steel using chemical composition and cooling rate.Then the SH-CCT diagrams of 6 kinds of steels are calculated by the hybrid machine learning model.The results show that the accuracy of the classification model is up to 100%,the predicted values of the regression models are in good agreement with the experimental results,with high correlation coefficient and low error value.Moreover,the mathematical expressions of hardness in welding heat-affected zone of low alloy steel are calculated by symbolic regression,which can quantitatively express the relationship between alloy composition,cooling time and hardness.This study demonstrates the great potential of the material informatics in the field of welding technology.
文摘The predominant method for smart phone accessing is confined to methods directing the authentication by means of Point-of-Entry that heavily depend on physiological biometrics like,fingerprint or face.Implicit continuous authentication initiating to be loftier to conventional authentication mechanisms by continuously confirming users’identities on continuing basis and mark the instant at which an illegitimate hacker grasps dominance of the session.However,divergent issues remain unaddressed.This research aims to investigate the power of Deep Reinforcement Learning technique to implicit continuous authentication for mobile devices using a method called,Gaussian Weighted Cauchy Kriging-based Continuous Czekanowski’s(GWCK-CC).First,a Gaussian Weighted Non-local Mean Filter Preprocessing model is applied for reducing the noise pre-sent in the raw input face images.Cauchy Kriging Regression function is employed to reduce the dimensionality.Finally,Continuous Czekanowski’s Clas-sification is utilized for proficient classification between the genuine user and attacker.By this way,the proposed GWCK-CC method achieves accurate authen-tication with minimum error rate and time.Experimental assessment of the pro-posed GWCK-CC method and existing methods are carried out with different factors by using UMDAA-02 Face Dataset.The results confirm that the proposed GWCK-CC method enhances authentication accuracy,by 9%,reduces the authen-tication time,and error rate by 44%,and 43%as compared to the existing methods.
基金This study was supported in part by the Ministry of Science and Technology MOST 108-2221-E-150-022-MY3 and Taiwan Ocean University.
文摘This study proposed a measurement platform for continuous blood pressure estimation based on dual photoplethysmography(PPG)sensors and a deep learning(DL)that can be used for continuous and rapid measurement of blood pressure and analysis of cardiovascular-related indicators.The proposed platform measured the signal changes in PPG and converted them into physiological indicators,such as pulse transit time(PTT),pulse wave velocity(PWV),perfusion index(PI)and heart rate(HR);these indicators were then fed into the DL to calculate blood pressure.The hardware of the experiment comprised 2 PPG components(i.e.,Raspberry Pi 3 Model B and analog-todigital converter[MCP3008]),which were connected using a serial peripheral interface.The DL algorithm converted the stable dual PPG signals acquired from the strictly standardized experimental process into various physiological indicators as input parameters and finally obtained the systolic blood pressure(SBP),diastolic blood pressure(DBP)and mean arterial pressure(MAP).To increase the robustness of the DL model,this study input data of 100 Asian participants into the training database,including those with and without cardiovascular disease,each with a proportion of approximately 50%.The experimental results revealed that the mean absolute error and standard deviation of SBP was 0.17±0.46 mmHg.The mean absolute error and standard deviation of DBP was 0.27±0.52 mmHg.The mean absolute error and standard deviation of MAP was 0.16±0.40 mmHg.
文摘The overall research in Reinforcement Learning (RL) concentrates on discrete sets of actions, but for certain real-world problems it is important to have methods which are able to find good strategies using actions drawn from continuous sets. This paper describes a simple control task called direction finder and its known optimal solution for both discrete and continuous actions. It allows for comparison of RL solution methods based on their value functions. In order to solve the control task for continuous actions, a simple idea for generalising them by means of feature vectors is presented. The resulting algorithm is applied using different choices of feature calculations. For comparing their performance a simple measure is
文摘Cooperative learning emerging as the leading new approach to classroom instruction abroad over the past decades has been studied by many researchers from all aspects.This paper mainly focuses on the basics of cooperative learning and tries to answer the question that if the use of cooperative learning produce higher achievement than the traditional methods in college English reading class through experimental study.The analysis contributes to better college English teaching and learning.A conclusion is drawn that cooperative learning is very effective in improving college students reading ability.
