针对复杂交通场景车辆检测算法自适应能力差的问题,提出了基于Co-training半监督学习方法的车辆鲁棒检测算法.首先,针对手工标记的少量样本,分别训练基于Haar-like特征的AdaBoost分类器和基于HOG(histograms of oriented gradients)特征...针对复杂交通场景车辆检测算法自适应能力差的问题,提出了基于Co-training半监督学习方法的车辆鲁棒检测算法.首先,针对手工标记的少量样本,分别训练基于Haar-like特征的AdaBoost分类器和基于HOG(histograms of oriented gradients)特征的SVM(support vector machines)分类器,使其具有一定的识别能力;然后,基于Co-training半监督学习框架,将利用2种算法进行分类得到的新样本分别加入到对方的样本库中,增加训练样本数量,再次进行分类器的训练.由于这2类特征具有冗余性,各自检测出的正负样本包含对方漏检和误检的图像.由于样本数的增加,再次训练所得到的新分类器的鲁棒性得到了很大提高,能更加准确地检测出车辆,而且由算法对未标记样本进行分类标记,不再需要人为标记,提高了车辆检测算法的自适应能力.展开更多
Walking is the most basic and essential part of the activities of daily living. To enable the elderly and non-ambulatory gait-impaired patients, the repetitive practice of this task, a novel gait training robot(GTR) w...Walking is the most basic and essential part of the activities of daily living. To enable the elderly and non-ambulatory gait-impaired patients, the repetitive practice of this task, a novel gait training robot(GTR) was designed followed the end-effector principle, and an active partial body weight support(PBWS) system was introduced to facilitate successful gait training. For successful establishment of a walking gait on the GTR with PBWS, the motion laws of the GTR were planned to enable the phase distribution relationships of the cycle step, and the center of gravity(COG) trajectory of the human body during gait training on the GTR was measured. A coordinated control strategy was proposed based on the impedance control principle. A robotic prototype was developed as a platform for evaluating the design concepts and control strategies. Preliminary gait training with a healthy subject was implemented by the robotic-assisted gait training system and the experimental results are encouraging.展开更多
针对Tri-training算法利用无标记样例时会引入噪声且限制无标记样例的利用率而导致分类性能下降的缺点,提出了AR-Tri-training(Tri-training with assistant and rich strategy)算法。提出辅助学习策略,结合富信息策略设计辅助学习器,...针对Tri-training算法利用无标记样例时会引入噪声且限制无标记样例的利用率而导致分类性能下降的缺点,提出了AR-Tri-training(Tri-training with assistant and rich strategy)算法。提出辅助学习策略,结合富信息策略设计辅助学习器,并将辅助学习器应用在Tri-training训练以及说话声识别中。实验结果表明,辅助学习器在Tri-training训练的基础上不仅降低每次迭代可能产生的误标记样例数,而且能够充分地利用无标记样例以及在验证集上的错分样例信息。从实验结果可以得出,该算法能够弥补Tri-training算法的缺点,进一步提高测试率。展开更多
In engineering application,there is only one adaptive weights estimated by most of traditional early warning radars for adaptive interference suppression in a pulse reputation interval(PRI).Therefore,if the training s...In engineering application,there is only one adaptive weights estimated by most of traditional early warning radars for adaptive interference suppression in a pulse reputation interval(PRI).Therefore,if the training samples used to calculate the weight vector does not contain the jamming,then the jamming cannot be removed by adaptive spatial filtering.If the weight vector is constantly updated in the range dimension,the training data may contain target echo signals,resulting in signal cancellation effect.To cope with the situation that the training samples are contaminated by target signal,an iterative training sample selection method based on non-homogeneous detector(NHD)is proposed in this paper for updating the weight vector in entire range dimension.The principle is presented,and the validity is proven by simulation results.展开更多
In order to mitigate the effects of space adaptation syndrome(SAS) and improve the training efficiency of the astronauts, a novel astronaut rehabilitative training robot(ART) was proposed. ART can help the astronauts ...In order to mitigate the effects of space adaptation syndrome(SAS) and improve the training efficiency of the astronauts, a novel astronaut rehabilitative training robot(ART) was proposed. ART can help the astronauts to carry out the bench press training in the microgravity environment. Firstly, a dynamic model of cable driven unit(CDU) was established whose accuracy was verified through the model identification. Secondly, to improve the accuracy and the speed of the active loading, an active loading hybrid force controller was proposed on the basis of the dynamic model of the CDU. Finally, the actual effect of the hybrid force controller was tested by simulations and experiments. The results suggest that the hybrid force controller can significantly improve the precision and the dynamic performance of the active loading with the maximum phase lag of the active loading being 9° and the maximum amplitude error being 2% at the frequency range of 10 Hz. The controller can meet the design requirements.展开更多
In view of the uncertainty and complexity,the intelligent model of rehabilitation training program for stroke was proposed,combining with the case-based reasoning(CBR) and interval type-2 fuzzy reasoning(IT2FR).The mo...In view of the uncertainty and complexity,the intelligent model of rehabilitation training program for stroke was proposed,combining with the case-based reasoning(CBR) and interval type-2 fuzzy reasoning(IT2FR).The model consists of two parts:the setting model based on CBR and the feedback compensation model based on IT2FR.The former presets the value of rehabilitation training program,and the latter carries on the feedback compensation of the preset value.Experimental results show that the average percentage error of two rehabilitation training programs is 0.074%.The two programs are made by the intelligent model and rehabilitation physician.That is,the two different programs are nearly identical.It means that the intelligent model can make a rehabilitation training program effectively and improve the rehabilitation efficiency.展开更多
The optimal design of training sequences for channel estimation in multiple-input multiple-output （MIMO） systems under spatially correlated fading is considered. The channel is assumed to be a block-fading model wit...The optimal design of training sequences for channel estimation in multiple-input multiple-output （MIMO） systems under spatially correlated fading is considered. The channel is assumed to be a block-fading model with spatial correlation known at both the transmitter and the receiver. To minimize the channel estimation error, optimal training sequences are designed to exploit full information of the spatial correlation under the criterion of minimum mean square error （MMSE）. It is investigated that the spatial correlation is helpful to decrease the estimation error and the proposed training sequences have good performance via simulations.展开更多
Deep neural networks(DNNs)have achieved great success in many data processing applications.However,high computational complexity and storage cost make deep learning difficult to be used on resource-constrained devices...Deep neural networks(DNNs)have achieved great success in many data processing applications.However,high computational complexity and storage cost make deep learning difficult to be used on resource-constrained devices,and it is not environmental-friendly with much power cost.In this paper,we focus on low-rank optimization for efficient deep learning techniques.In the space domain,DNNs are compressed by low rank approximation of the network parameters,which directly reduces the storage requirement with a smaller number of network parameters.In the time domain,the network parameters can be trained in a few subspaces,which enables efficient training for fast convergence.The model compression in the spatial domain is summarized into three categories as pre-train,pre-set,and compression-aware methods,respectively.With a series of integrable techniques discussed,such as sparse pruning,quantization,and entropy coding,we can ensemble them in an integration framework with lower computational complexity and storage.In addition to summary of recent technical advances,we have two findings for motivating future works.One is that the effective rank,derived from the Shannon entropy of the normalized singular values,outperforms other conventional sparse measures such as the?_1 norm for network compression.The other is a spatial and temporal balance for tensorized neural networks.For accelerating the training of tensorized neural networks,it is crucial to leverage redundancy for both model compression and subspace training.展开更多
美国生物安全培训课程（Biosafety and Biosecurity Training Course,BBTC）是在美国科罗拉多州柯林斯堡举办的为期八天的一个密集培训班。BBTC由科罗拉多州立大学生物安全主任Robert Ellis博士发起并担任主任。Ellis博士同时兼任兽医...美国生物安全培训课程（Biosafety and Biosecurity Training Course,BBTC）是在美国科罗拉多州柯林斯堡举办的为期八天的一个密集培训班。BBTC由科罗拉多州立大学生物安全主任Robert Ellis博士发起并担任主任。Ellis博士同时兼任兽医和生物医学学院教授,是在美国颇有影响的生物安全专家,具有多年的生物安全实践和管理经验。展开更多
The paper introduces some technology for training, simulation, restoration expert system of power grid, the structure of the system including function composition, hardware and software composition are discussed, know...The paper introduces some technology for training, simulation, restoration expert system of power grid, the structure of the system including function composition, hardware and software composition are discussed, knowledge representation and the method to establish device graphical library for expert system are given, the fault setting and diagnosis for training and simulation as well as restoration technology with deep first searching arithmetic and heuristic inference are presented. The research provides a good base for developing the training, simulation, restoration system of power companies.展开更多
With the development of computer science,the software technology changes with each passing day,itput forward higher and higher technical requirements for the software developers.Aim at each link in the process of deve...With the development of computer science,the software technology changes with each passing day,itput forward higher and higher technical requirements for the software developers.Aim at each link in the process of development,the present paper put forward a kind of software engineering personnel training system,the system can build a unified learning,management and evaluation system,to avoid the disadvantages of the traditional single system structure;the training process is more flexible,and it couldreduce the complexity of the artificial training and cause it to become more practical value.展开更多
基金Project(61175128) supported by the National Natural Science Foundation of ChinaProject(2008AA040203) supported by the National High Technology Research and Development Program of China
文摘Walking is the most basic and essential part of the activities of daily living. To enable the elderly and non-ambulatory gait-impaired patients, the repetitive practice of this task, a novel gait training robot(GTR) was designed followed the end-effector principle, and an active partial body weight support(PBWS) system was introduced to facilitate successful gait training. For successful establishment of a walking gait on the GTR with PBWS, the motion laws of the GTR were planned to enable the phase distribution relationships of the cycle step, and the center of gravity(COG) trajectory of the human body during gait training on the GTR was measured. A coordinated control strategy was proposed based on the impedance control principle. A robotic prototype was developed as a platform for evaluating the design concepts and control strategies. Preliminary gait training with a healthy subject was implemented by the robotic-assisted gait training system and the experimental results are encouraging.
文摘针对Tri-training算法利用无标记样例时会引入噪声且限制无标记样例的利用率而导致分类性能下降的缺点,提出了AR-Tri-training(Tri-training with assistant and rich strategy)算法。提出辅助学习策略,结合富信息策略设计辅助学习器,并将辅助学习器应用在Tri-training训练以及说话声识别中。实验结果表明,辅助学习器在Tri-training训练的基础上不仅降低每次迭代可能产生的误标记样例数,而且能够充分地利用无标记样例以及在验证集上的错分样例信息。从实验结果可以得出,该算法能够弥补Tri-training算法的缺点,进一步提高测试率。
基金supported by the National Natural Science Foundation of China(62371049)。
文摘In engineering application,there is only one adaptive weights estimated by most of traditional early warning radars for adaptive interference suppression in a pulse reputation interval(PRI).Therefore,if the training samples used to calculate the weight vector does not contain the jamming,then the jamming cannot be removed by adaptive spatial filtering.If the weight vector is constantly updated in the range dimension,the training data may contain target echo signals,resulting in signal cancellation effect.To cope with the situation that the training samples are contaminated by target signal,an iterative training sample selection method based on non-homogeneous detector(NHD)is proposed in this paper for updating the weight vector in entire range dimension.The principle is presented,and the validity is proven by simulation results.
基金Project(61175128) supported by the National Natural Science Foundation of ChinaProject(2008AA040203) supported by the National High Technology Research and Development Program of ChinaProject(QC2010009) supported by the Natural Science Foundation of Heilongjiang Province,China
文摘In order to mitigate the effects of space adaptation syndrome(SAS) and improve the training efficiency of the astronauts, a novel astronaut rehabilitative training robot(ART) was proposed. ART can help the astronauts to carry out the bench press training in the microgravity environment. Firstly, a dynamic model of cable driven unit(CDU) was established whose accuracy was verified through the model identification. Secondly, to improve the accuracy and the speed of the active loading, an active loading hybrid force controller was proposed on the basis of the dynamic model of the CDU. Finally, the actual effect of the hybrid force controller was tested by simulations and experiments. The results suggest that the hybrid force controller can significantly improve the precision and the dynamic performance of the active loading with the maximum phase lag of the active loading being 9° and the maximum amplitude error being 2% at the frequency range of 10 Hz. The controller can meet the design requirements.
基金Project(2010020176-301)supported by Liaoning Science and Technology Program,ChinaProject(F10-2D5-1-57)supported by Shenyang Municipal Fund,China
文摘In view of the uncertainty and complexity,the intelligent model of rehabilitation training program for stroke was proposed,combining with the case-based reasoning(CBR) and interval type-2 fuzzy reasoning(IT2FR).The model consists of two parts:the setting model based on CBR and the feedback compensation model based on IT2FR.The former presets the value of rehabilitation training program,and the latter carries on the feedback compensation of the preset value.Experimental results show that the average percentage error of two rehabilitation training programs is 0.074%.The two programs are made by the intelligent model and rehabilitation physician.That is,the two different programs are nearly identical.It means that the intelligent model can make a rehabilitation training program effectively and improve the rehabilitation efficiency.
基金the National Science Foundation for Distinguished Young Scholars (60725105)the SixthProject of the Key Project of National Nature Science Foundation of China (60496316)+2 种基金the National "863" Project (2007AA012288)the National Nature Science Foundation of China (60572146)the "111" Project (B08038).
文摘The optimal design of training sequences for channel estimation in multiple-input multiple-output （MIMO） systems under spatially correlated fading is considered. The channel is assumed to be a block-fading model with spatial correlation known at both the transmitter and the receiver. To minimize the channel estimation error, optimal training sequences are designed to exploit full information of the spatial correlation under the criterion of minimum mean square error （MMSE）. It is investigated that the spatial correlation is helpful to decrease the estimation error and the proposed training sequences have good performance via simulations.
基金supported by the National Natural Science Foundation of China(62171088,U19A2052,62020106011)the Medico-Engineering Cooperation Funds from University of Electronic Science and Technology of China(ZYGX2021YGLH215,ZYGX2022YGRH005)。
文摘Deep neural networks(DNNs)have achieved great success in many data processing applications.However,high computational complexity and storage cost make deep learning difficult to be used on resource-constrained devices,and it is not environmental-friendly with much power cost.In this paper,we focus on low-rank optimization for efficient deep learning techniques.In the space domain,DNNs are compressed by low rank approximation of the network parameters,which directly reduces the storage requirement with a smaller number of network parameters.In the time domain,the network parameters can be trained in a few subspaces,which enables efficient training for fast convergence.The model compression in the spatial domain is summarized into three categories as pre-train,pre-set,and compression-aware methods,respectively.With a series of integrable techniques discussed,such as sparse pruning,quantization,and entropy coding,we can ensemble them in an integration framework with lower computational complexity and storage.In addition to summary of recent technical advances,we have two findings for motivating future works.One is that the effective rank,derived from the Shannon entropy of the normalized singular values,outperforms other conventional sparse measures such as the?_1 norm for network compression.The other is a spatial and temporal balance for tensorized neural networks.For accelerating the training of tensorized neural networks,it is crucial to leverage redundancy for both model compression and subspace training.
文摘美国生物安全培训课程（Biosafety and Biosecurity Training Course,BBTC）是在美国科罗拉多州柯林斯堡举办的为期八天的一个密集培训班。BBTC由科罗拉多州立大学生物安全主任Robert Ellis博士发起并担任主任。Ellis博士同时兼任兽医和生物医学学院教授,是在美国颇有影响的生物安全专家,具有多年的生物安全实践和管理经验。
基金TheKeyProblemTacklingProjectinHunanProvince! (No .Izf 9831)
文摘The paper introduces some technology for training, simulation, restoration expert system of power grid, the structure of the system including function composition, hardware and software composition are discussed, knowledge representation and the method to establish device graphical library for expert system are given, the fault setting and diagnosis for training and simulation as well as restoration technology with deep first searching arithmetic and heuristic inference are presented. The research provides a good base for developing the training, simulation, restoration system of power companies.
文摘With the development of computer science,the software technology changes with each passing day,itput forward higher and higher technical requirements for the software developers.Aim at each link in the process of development,the present paper put forward a kind of software engineering personnel training system,the system can build a unified learning,management and evaluation system,to avoid the disadvantages of the traditional single system structure;the training process is more flexible,and it couldreduce the complexity of the artificial training and cause it to become more practical value.
