Journal of Intelligent & Fuzzy Systems - Volume 38, issue 1

Authors: Zhang, Xinsheng | Gao, Teng

Article Type: Research Article

Abstract: Aspect level sentiment classification task requires topical polarity classification for different description aspect. There is a polysemy in the same vocabulary, and the emotional polarity is different for different objects. Word embedding can capture semantic information but cannot adapt to the polysemy. Attention mechanism has achieved good performance in the above tasks; however, it is only able to get the degree of association between words and unable to get detailed descriptions. In this paper, the ELMOs model is used to adjust the polysemy of the word. The Transformer model is used to extract the features with the highest degree of …relevance to the target object for emotional polarity classification. Our work contribution is to overcome the polysemy interference, and use the attention mechanism to model the network relationship between words, so that the model can extract important classification features according to different target words. Experiments on laptop and restaurant datasets demonstrate that our approach achieves a new state-of-the-art performance on a few benchmarks. Show more

Keywords: Text sentiment classification, fine-grained sentiment analysis, attention mechanism

DOI: 10.3233/JIFS-179383

Citation: Journal of Intelligent & Fuzzy Systems, vol. 38, no. 1, pp. 89-96, 2020

Authors: Wang, Zhongru | Ruan, Qiang

Article Type: Research Article

Abstract: Communication network security is an important part of the digital signal processor. In particular, the process bus replaces the traditional hard wiring, which makes the network system extend from the second to the first time, greatly increasing the network scale and network traffic; the information transmitted on the process bus is absolutely large. Most requirements require strict real-time and high reliability. Therefore, information security is a major problem that threatens the security, stability, economy, and quality operation of network systems, and needs to be paid enough attention. This paper introduces the methods of information classification and information merging to improve …the real-time information. It proposes to apply network security technologies such as information encryption technology, firewall technology, mobile agent, security management technology and virtual private network (VPN) technology to office network of the electric-power industry and analysed the specific application scenarios and effects. Show more

Keywords: Network video surveillance, embedded system, digital signal processor, network security subsystem

DOI: 10.3233/JIFS-179384

Citation: Journal of Intelligent & Fuzzy Systems, vol. 38, no. 1, pp. 97-103, 2020

Authors: Zang, Jingfeng | Ren, Guibin | An, Yanlin | Piao, Yan

Article Type: Research Article

Abstract: Bad weather has a negative effect on the perceptual quality and degrades the performance of computer vision system. Therefore, a rain removal method based on dual-tree complex wavelet fusion is proposed. The algorithm can be further used for video surveillance system and intelligent transportation and other fields. The method analyzes from the perspective of frequency domain, using the dual-tree complex wavelet decomposition: decomposing images into low frequency sub-images and high frequency sub-images, then developing the different fusion rules. For the low frequency sub-images, fusion rules using the principal component analysis. For the high frequency sub-images, fusion rules using the local …energy matching. In this paper, an image edge enhancement algorithm based on fast guided filter is proposed, a SIFT feature matching method based on maximum likelihood estimation sampling and consistent(MLESAC) algorithm is proposed. Experiments results show that the proposed algorithm can improve the definition of images and restore the details of the target blocked by raindrops. Show more

Keywords: Raindrops removal, dual-tree complex wavelet fusion, PCA, local energy

DOI: 10.3233/JIFS-179385

Citation: Journal of Intelligent & Fuzzy Systems, vol. 38, no. 1, pp. 105-113, 2020

Authors: Wang, Dan | Zhao, Hongwei | Li, Qingliang

Article Type: Research Article

Abstract: This paper designs a brand-new image retrieval method of mammary cancer based on convolution neural network. This method simulates VLAD layer in CNN network structure, designs a trainable universal VLAD layer-NET VLAD layer, reduces dimensions and optimizes VLAD descriptors, applies structure from motion algorithm to automatically label samples, and obtains the minimum loss function value by a new training program of weakly supervised ranking loss. Experiments show that this method has improved retrieval performance compared with similar retrieval methods and non-network structure retrieval methods.

Keywords: Convolutional neural network, VLAD, loss function, medical image retrieval

DOI: 10.3233/JIFS-179386

Citation: Journal of Intelligent & Fuzzy Systems, vol. 38, no. 1, pp. 115-126, 2020

Authors: Wang, Dan | Zhao, Hongwei | Li, Qingliang

Article Type: Research Article

Abstract: This paper presents a medical brain image algorithm based on multi-feature fusion. Feature extraction based on convolutional neural network was used as texture information, feature extraction based on voxel information was used as morphological feature, and then the two types of features were combined in series. Feature extraction based on convolutional neural network was used as texture information, feature extraction based on voxel information was used as morphological feature, and then the two types of features were combined in series. Then the heuristic search algorithm is used to optimize the feature selection stage. Based on the feature score table extracted …by the recursive feature elimination method of support vector machine, the correlation between features is added. Moreover, through experimental analysis, the optimal value of the parameter K was selected according to the heuristic search, and the optimal feature subset was extracted after determining the value of the parameter K. Experiments show that compared with similar algorithms, this algorithm improves the accuracy and efficiency of the classification of brain images. Show more

Keywords: Convolutional neural network, multi-feature fusion, heuristic search, medical image classification

DOI: 10.3233/IFS-179387

Citation: Journal of Intelligent & Fuzzy Systems, vol. 38, no. 1, pp. 127-137, 2020

Authors: Zeng, Jun

Article Type: Research Article

Abstract: In order to improve the security of information storage in multi-domain optical network privacy protection, hybrid encoding and encryption of privacy protection information in multi-domain optical network is carried out. A hybrid coding and secure encryption technique based on particle swarm optimization (PSO) for privacy protection information in multi-domain optical network is proposed. The coding mapping equation of privacy protection information security encryption in multi-domain optical network is constructed, and the privacy protection information of multi-domain fiber network is loaded into the coding mapping equation of security encryption of privacy protection information in multi-domain fiber network. A set of characteristic …solutions describing the entropy function of the characteristic distribution of random encryption keys are obtained, and the elliptic mapping random linear combination coding and chaotic encryption key allocation are carried out. The privacy protection method of multi-domain optical fiber network is used to encrypt and decode information with mixed coding and steganography. Under the bilinear mapping coding system, the privacy-protected information in multi-domain optical network is encrypted and encrypted, stored and transmitted confidently. The simulation results show that this encryption technique has better steganography performance and better secure transmission and storage performance for the privacy protection information mixed encoding encryption in multi-domain optical network. Show more

Keywords: Particle swarm optimization algorithm, multi-domain fiber optic network, secure encryption

DOI: 10.3233/JIFS-179388

Citation: Journal of Intelligent & Fuzzy Systems, vol. 38, no. 1, pp. 139-145, 2020

Authors: Chen, Limin | Li, Zhuohang | Lv, Muzhan | Xiong, Mingliang

Article Type: Research Article

Abstract: In order to improve the ability of automatic estimation and prediction of economic trend index, an intelligent prediction model of economic trend index based on rough set support vector machine is proposed. The statistical analysis of intelligent prediction of economic trend index is carried out by using the equivalent approximate linear model, and the regression analysis model of intelligent prediction of economic trend index is established. Combining with the rough set support vector machine big data fusion technology, the feature extraction and information mining are carried out in the process of intelligent prediction of economic trend index, and the statistical …time analysis series of economic trend index is constructed. The spatial distribution of economic trend index distribution series is reconstructed, and the economic trend is evaluated and predicted in the high dimensional economic trend index forecast series distribution space. The principal component characteristic analysis and fuzzy closeness analysis of economic trend index are carried out by using fuzzy relational degree scheduling method. Taking economic cost, economic development prospect and economic growth rate as constraint indexes, the method of multi-factor joint estimation is adopted. Realize economic trend index intelligent forecast. The simulation results show that the accuracy of fast estimation of economic trend index is high, the time cost is small, and the ability of intelligent prediction is stronger. Show more

Keywords: Rough set, support vector machine, economic trend index, intelligent prediction

DOI: 10.3233/JIFS-179389

Citation: Journal of Intelligent & Fuzzy Systems, vol. 38, no. 1, pp. 147-153, 2020

Authors: Sun, Xiang

Article Type: Research Article

Abstract: With the continuous progress of network technology, some abnormal data are often confused in network data flow, which affects network security. In order to grasp the abnormal degree of abnormal data in networks and detect the similarity of abnormal data, an optimized genetic data mining algorithm is used to mine abnormal data in network, obtain the initial population of abnormal data mining and optimize genetic operation. On this basis, the network data type and the number of network data types are adaptively adjusted to obtain the optimal abnormal data mining results. Based on Euclidean distance, the similarity value of abnormal …data in network is calculated, and the greater the similarity value is, the greater the similarity of abnormal data is and vice versa. The experimental results show that the average standard deviation of detection error and energy consumption of the proposed method are 0.00865 and 398J, respectively. This method is a reliable and energy-saving method for similarity detection of abnormal data in network, which provides an effective basis for grasping the anomaly degree of network data. Show more

Keywords: Data mining, abnormal data in network, population, optimized genetics, similarity, detection

DOI: 10.3233/JIFS-179390

Citation: Journal of Intelligent & Fuzzy Systems, vol. 38, no. 1, pp. 155-162, 2020

Authors: Li, Dakai | Zhang, Fan | Tian, Yuanli

Article Type: Research Article

Abstract: In the era of the Internet of Things, the way of information dissemination and the speed of information computing have changed dramatically. New generation technologies such as big data, IOTs, cloud computing, and mobile internet have rapidly developed, and the information resources in the process of using customers have promoted to value creation activities. The guiding force, therefore, the service model must also follow the changes. This paper is based on the enterprise management integration mechanism and information platform research of the Internet of Things environment, and analyses its conceptual model, the relationship between the platform and related subjects, the …value creation mechanism, the core business functions and the profit model, which have certain guiding significance for practice. The enterprise management integration mechanism and information platform research in the Internet of Things environment requires the parties to interact, share, and cooperate with each other to achieve common value creation. In addition, enterprises effectively research internal resource integration and information sharing, rationally streamline enterprise management institutions, thereby improving the quality and efficiency of business operations, giving full play to the role of internal control and risk management, so that enterprises can develop steadily. Show more

Keywords: Enterprise, information platform, internal control, Internet of Things, management integration

DOI: 10.3233/JIFS-179391

Citation: Journal of Intelligent & Fuzzy Systems, vol. 38, no. 1, pp. 163-173, 2020

Authors: Xu, Tian | Fan, Jizhuang | Fang, Qianqian | Zhu, Yanhe | Zhao, Jie

Article Type: Research Article

Abstract: Collision detection is the core issue in physical human–robot interactions, and many detection methods based on robot dynamic models have been proposed. However, model uncertainties, especially complicated friction, seriously affect the collision detection performance of these methods. In this paper, a nonlinear disturbance observer (NDO) originally proposed for friction estimation is applied for the first time in robot collision detection. To verify that the collision detection performance of the NDO is better than that of the classical generalized momentum observer (GMO), the detection sensitivity, robustness and external torque estimation accuracy of each method are compared and analyzed. Then, to eliminate …the effects of friction uncertainties on the collision detection results, a modified nonlinear disturbance observer (MNDO) based on neural networks is proposed to improve the collision detection performance. To verify the effectiveness of the algorithm, simulations and experiments are conducted with a 6-DOF robot and two single-joint platforms. The results indicate that the proposed algorithm is accurate and effective. Show more

Keywords: Robot collision detection, NDO, GMO, friction estimation, neural networks

DOI: 10.3233/JIFS-179392

Citation: Journal of Intelligent & Fuzzy Systems, vol. 38, no. 1, pp. 175-186, 2020