Time filter

Source Type

Yan A.,Beijing University of Technology | Yan A.,Beijing Key Laboratory of Computational Intelligence and Intelligent System | Wang D.,La Trobe University
Expert Systems with Applications | Year: 2015

To achieve better classification performance using case-based reasoning classifiers, we propose a retrieval-based revision method with trustworthiness evaluation for problem solving. An improved case evaluation method is employed to evaluate the trustworthiness of the suggested solution after the reuse step, which will divide the target cases and its suggested solutions into a trustworthy set and an untrustworthy set in accordance with a threshold value of trustworthiness. The attribute weights are adjusted by running a genetic algorithm and are used in the second round of retrieval of the untrustworthy set to obtain the classification results. Experimental results demonstrate that our proposed method performs favorably compared with other methods. Also, the proposed method has less computation complexity for the trustworthiness evaluation, and enhances understanding on thinking and inference for case-based reasoning classifiers. © 2015 Elsevier Ltd. All rights reserved.


Li W.,Beijing University of Technology | Li W.,CAS Institute of Automation | Li W.,Beijing Key Laboratory of Computational Intelligence and Intelligent System | Li J.,Capital Medical University | And 5 more authors.
Neural Plasticity | Year: 2016

Previous neuroimaging studies suggested structural or functional brain reorganizations occurred in prelingually deaf subjects. However, little is known about the reorganizations of brain network architectures in prelingually deaf adolescents. The present study aims to investigate alterations of whole-brain functional network using resting-state fMRI and graph theory analysis. We recruited 16 prelingually deaf adolescents (1018 years) and 16 normal controls matched in age and gender. Brain networks were constructed from mean time courses of 90 regions. Widely distributed network was observed in deaf subjects, with increased connectivity between the limbic system and regions involved in visual and language processing, suggesting reinforcement of the processing for the visual and verbal information in deaf adolescents. Decreased connectivity was detected between the visual regions and language regions possibly due to inferior reading or speaking skills in deaf subjects. Using graph theory analysis, we demonstrated small-worldness property did not change in prelingually deaf adolescents relative to normal controls. However, compared with healthy adolescents, eight regions involved in visual, language, and auditory processing were identified as hubs only present in prelingually deaf adolescents. These findings revealed reorganization of brain functional networks occurred in prelingually deaf adolescents to adapt to deficient auditory input. © 2016 Wenjing Li et al.


Li W.,Beijing University of Technology | Li W.,CAS Institute of Automation | Li W.,Beijing Key Laboratory of Computational Intelligence and Intelligent System | Li J.,Capital Medical University | And 6 more authors.
Restorative Neurology and Neuroscience | Year: 2015

Purpose: Previous studies have shown brain reorganizations after early deprivation of auditory sensory. However, changes of grey matter connectivity have not been investigated in prelingually deaf adolescents yet. In the present study, we aimed to investigate changes of grey matter connectivity within and between auditory, language and visual systems in prelingually deaf adolescents. Methods:We recruited 16 prelingually deaf adolescents and 16 age-and gender-matched normal controls, and extracted the grey matter volume as the structural characteristic from 14 regions of interest involved in auditory, language or visual processing to investigate the changes of grey matter connectivity within and between auditory, language and visual systems. Sparse inverse covariance estimation (SICE) was utilized to construct grey matter connectivity between these brain regions. Results: The results show that prelingually deaf adolescents present weaker grey matter connectivity within auditory and visual systems, and connectivity between language and visual systems declined. Notably, significantly increased brain connectivity was found between auditory and visual systems in prelingually deaf adolescents. Conclusions: Our results indicate "cross-modal" plasticity after deprivation of the auditory input in prelingually deaf adolescents, especially between auditory and visual systems. Besides, auditory deprivation and visual deficits might affect the connectivity pattern within language and visual systems in prelingually deaf adolescents. © 2015-IOS Press and the authors. All rights reserved.


Qiao J.,Beijing University of Technology | Qiao J.,Beijing Key Laboratory of Computational Intelligence and Intelligent System | Li W.,Beijing University of Technology | Li W.,Beijing Key Laboratory of Computational Intelligence and Intelligent System | And 3 more authors.
Applied Soft Computing Journal | Year: 2016

This paper proposes a two-phase identification approach to Mamdani fuzzy neural networks. The first phase is the system identification which includes a novel forward recursive input-output clustering method for the structure initialization and the gradient descent algorithm for the parameter initialization. The main advantage of the proposed method is that it fits perfectly the special clustering requirement for system identification: coarser clustering in the regions where the identified system is smoother and finer clustering in the regions where the system is more variable or nonlinear. The second phase is the system simplification which includes the accurate similarity analysis and merging method for similar fuzzy rules and the gradient descent algorithm for the parameter finalization. The accurate similarity analysis developed solves the long standing open problem how to compute the exact (rather than approximate) similarity between fuzzy sets and rules with Gaussian membership functions. Numerical experiments based on well-known benchmark data sets are used to verify the effectiveness and accuracy of the proposed approach. © 2016 Elsevier B.V.


Yang G.,Beijing University of Technology | Yang G.,East China Jiaotong University | Qiao J.F.,Beijing University of Technology | Qiao J.F.,Beijing Key Laboratory of Computational Intelligence and Intelligent System
Applied Soft Computing Journal | Year: 2014

Spatial architecture neural network (SANN), which is inspired by the connecting mode of excitatory pyramidal neurons and inhibitory interneurons of neocortex, is a multilayer artificial neural network and has good learning accuracy and generalization ability when used in real applications. However, the backpropagation-based learning algorithm (named BP-SANN) may be time consumption and slow convergence. In this paper, a new fast and accurate two-phase sequential learning scheme for SANN is hereby introduced to guarantee the network performance. With this new learning approach (named SFSL-SANN), only the weights connecting to output neurons will be trained during the learning process. In the first phase, a least-squares method is applied to estimate the span-output-weight on the basis of the fixed randomly generated initialized weight values. The improved iterative learning algorithm is then used to learn the feedforward-output-weight in the second phase. Detailed effectiveness comparison of SFSL-SANN is done with BP-SANN and other popular neural network approaches on benchmark problems drawn from the classification, regression and time-series prediction applications. The results demonstrate that the SFSL-SANN is faster convergence and time-saving than BP-SANN, and produces better learning accuracy and generalization performance than other approaches. © 2014 Elsevier B.V. All rights reserved.


Zuo G.,Beijing University of Technology | Zuo G.,Beijing Key Laboratory of Computational Intelligence and Intelligent System | Liu Y.,Beijing University of Technology | Wang X.,Beijing University of Technology
Proceedings of the World Congress on Intelligent Control and Automation (WCICA) | Year: 2016

This paper presents a design of a bionic robot inspired by the hopping locomotion of kangaroo including the hopping and transmission mechanisms. The jumping movement can be realized by the hopping mechanism while the transmission mechanism is used to transmit energy from the power source to the spring and release energy when necessary. The kinematic equations of the robot at take-off phase are established. Characteristics of the displacement and the velocity of the center of mass and supporting force during take-off are analyzed as well as the energy conversion efficiency of the robot. The simulation results show that the mechanical structure of the kangaroo-bionic hopping robot is well designed to meet the bionic hopping characteristics and has a high energy conversion efficiency. © 2016 IEEE.


Li Z.,Beijing University of Technology | Zhang F.,Beijing University of Technology | Sun G.,Beijing University of Technology | Sun G.,Beijing Key Laboratory of Computational Intelligence and Intelligent System | And 2 more authors.
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) | Year: 2016

Due to non-invasiveness, monitoring driver state in computer vision (CV) has become a major way to detect driver fatigue. In contrast to other researches, we brought the driver fatigue detection system on the basis of DSP platform, which can make contribution to application on the integrated system for vehicle. However, the conventional system cannot easily be transplanted into DSP due to its storage and computation capacity. Therefore, designing an algorithm that can detect the fatigue efficiently is goal in this study. As the most important part of system, the geometric relationship and shape information within near frontal face is employed in the eye detection part, which depicts the eyebrow, eye and nose. In experiment part, a self-made database is assembled to test the performance of system. As the results of experiment, the detection rate of eye is achieved at 92.71% and driver fatigue state is obtained at 97.5%. © Springer International Publishing Switzerland 2016.


Pei F.,Beijing University of Technology | Pei F.,Beijing Key Laboratory of Computational Intelligence and Intelligent System | Zhu L.,Beijing University of Technology | Zhu L.,Beijing Key Laboratory of Computational Intelligence and Intelligent System | And 2 more authors.
Mathematical Problems in Engineering | Year: 2015

The accurate initial attitude is essential to affect the navigation result of Rotary Strapdown Inertial Navigation System (SINS), which is usually calculated by initial alignment. But marine mooring Rotary SINS has to withstand dynamic disturbance, such as the interference angular velocities and accelerations caused by surge and sway. In order to overcome the limit of dynamic disturbance under the marine mooring condition, an alignment method using novel adaptive Kalman filter for marine mooring Rotary SINS is developed in this paper. This alignment method using the gravity in the inertial frame as a reference is discussed to deal with the lineal and angular disturbances. Secondly, the system error model for fine alignment in the inertial frame as a reference is established. Thirdly, PWCS and SVD are used to analyze the observability of the system error model for fine alignment. Finally, a novel adaptive Kalman filter with measurement residual to estimate measurement noise variance is designed. The simulation results demonstrate that the proposed method can achieve better accuracy and stability for marine Rotary SINS. © 2015 Fujun Pei et al.


Qiao J.,Beijing University of Technology | Qiao J.,Beijing Key Laboratory of Computational Intelligence and Intelligent System | Li S.,Beijing University of Technology | Li S.,Beijing Key Laboratory of Computational Intelligence and Intelligent System | And 2 more authors.
Neurocomputing | Year: 2016

When a sigmoidal feedforward neural network (SFNN) is trained by the gradient-based algorithms, the quality of the overall learning process strongly depends on the initial weights. To improve the algorithm stability and avoid local minima, a Mutual Information based weight initialization (MIWI) method is proposed for SFNN. The useful information contained in input variables is measured with the mutual information (MI) between input variables and output variables. The initial distribution of weights is consistent with the information distribution in the input variables. The lower and upper bounds of the weights range are calculated to ensure the neurons inputs are within the active region of sigmoid function. The MIWI method makes the initial weights close to the global optimal point with a higher probability and avoids premature saturation. The efficiency of the MIWI method is evaluated based on several benchmark problems. The experimental results show that the stability and accuracy of the proposed method are better than some other weight initialization methods. © 2016 Elsevier B.V.


Zuo G.,Beijing University of Technology | Zuo G.,Beijing Key Laboratory of Computational Intelligence and Intelligent System | Wu C.,Beijing University of Technology
Proceedings of the 28th Chinese Control and Decision Conference, CCDC 2016 | Year: 2016

This paper proposes a heuristic Monte Carlo Tree Search (MCTS) method for a computer game, Surakarta chess. The proposed method uses heuristic knowledge to guide the search procedure heading to a better solution. In our implementation we use a CPU based root parallelization technology which permits many instances running simultaneously. The reliability of our method has been proved to be considerably effective by experimental results. © 2016 IEEE.

Loading Beijing Key Laboratory of Computational Intelligence and Intelligent System collaborators
Loading Beijing Key Laboratory of Computational Intelligence and Intelligent System collaborators