Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2022-05-10 Cooperative journals: 《计算机应用研究》
Abstract： Analyzing the evaluation data of students to teachers in the teaching evaluation system helps teachers understand the true attitudes of students to teachers, summarize teaching experience, improve subsequent teaching methods, and improve teaching quality. However, when evaluating teaching, random or malicious evaluations may occur among students, resulting in a large amount of noise in the evaluation data, which results in unsatisfactory feedback data. Therefore, this paper proposes a discrete Poisson mixture model to model the evaluation data of students with noise. Each discrete Poisson component in the mixture model corresponds to a class of students with similar evaluation modes. The model parameters in the loose distribution represent the evaluation scores in the corresponding evaluation mode. The log-likelihood function is constructed to measure the degree of fit between the mixed model and the evaluation data, and the gradient descent method is used to solve the model parameters with the highest degree of fit, to find the true evaluation of the students to the teacher, and to ensure the teacher-student relationship in the teaching evaluation system Communicate effectively. A large number of experimental results show that the model in this paper can quickly and accurately identify students with different evaluation modes from the evaluation data containing noise, and grasp the true evaluation of the students to teachers.
Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2019-05-10 Cooperative journals: 《计算机应用研究》
Abstract： Tag collision is one of the most common and difficult problems in RFID system. Solving the collision problem and reducing the tag recognition time are of great significance for the application of RFID. In order to solve the problem that some existing anti-collision algorithms were mostly applied in tag fixed scene but had poor performance in tag moving scene, this paper proposed an anti-collision algorithm in tag moving scene (TMS) . The algorithm first distinguished between the move-in tag and the resident tag, and then estimated the number of tags, finally used a hybrid recognition strategy to identify the tag based on estimated number of tags. The simulation results show that compared with other algorithms, TMS algorithm can effectively reduce the tag recognition time in tag moving scene, which has certain significance for the research of RFID tag anti-collision algorithm.
Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2018-07-09 Cooperative journals: 《计算机应用研究》
Abstract： Chinese word segmentation is one of the key technologies of Chinese natural language processing tasks. The existing neural network models based on deep learning are usually trained on single criterion corpora. This paper proposed a joint learning method based on bi-directional long short-term memory neural network and Conditional Random Fields for large-scale corpora. The corpora were composed of simplified Chinese data sets (PKU, MSRA, CTB6) and traditional Chinese data sets (CITYU, MSR) . A pair of identifiers is added to the beginning and end of each input sentence of the data set. The results of the experiments show that the effective method has good effect on Chinese word segmentation for such large-scale corpora.
Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2018-05-24 Cooperative journals: 《计算机应用研究》
Abstract： In order to effectively identify the sentiment polarity of Chinese microblogs, based on the cognitive fact that emojis can change or enhance the sentiment polarity of the texts, this paper proposed an emoji attentional neural network model for microblog sentiment analysis. This model firstly use a Bi-LSTM model to learn the feature representation of the text, and then an emoji-based attention mechanism is implemented to obtain a new feature representation after combining text with the emojis, so as to recognize the sentiment polarity of the microblogs. The experiments show that compared with the Bi-LSTM model that inputs plain text and emojis, the accuracy rate of the emoji-attentional model is increased by 4.06%; compared with the Bi-LSTM model that only inputs plain text, the accuracy rate of emoji-attentional model is increased by 6.35%.
Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2018-05-02 Cooperative journals: 《计算机应用研究》
Abstract： Given the sensitivity of human face angle, expression and attitude as well as the low recognition precision, this paper presented a new face recognition algorithm based on distance optimization method. There were two revised points in the proposed method: a) The algorithm used the LBP operator to extract the texture map of the face image and then converged it with the R, G and B channels of the original image, then the neural network could take the fused image matrix as input, which enriched the human face texture features; b) In the processing of training, it reconstructed the loss function to improve the performance, made use of the thresold and margin to constrain the distance of the feature vector, and built a new optimization target for the model, which could limit the faces of the same person have small euclidean distances and faces of distinct people have large distances. Experiments on the unconstrained scenes of the LFW face database show that the accuracy of the model is 99.15% respectively, indict the model can improve the accuracy of face recognition with strong robustness.
Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2018-04-19 Cooperative journals: 《计算机应用研究》
Abstract： For the low accuracy of earthquake prediction caused by the lack of large earthquake data, this paper proposed an earthquake prediction method based on negative selection. In this method, it used the variable real valued negative selection algorithm to generate mature detectors which used to predict whether an earthquake occured. Due to the absence of non-self data sets in the negative selection training, it could be reduced the impact of lacking large earthquake data on the training effect. It used the historical earthquake data of Sichuan province to predict whether the magnitude 5.0 and above earthquakes occurred in Sichuan within one month. Compared with the traditional machine learning algorithm, the results show that the negative selection has better prediction effect.
Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2018-04-17 Cooperative journals: 《计算机应用研究》
Abstract： Under the circumstance of insufficient prior knowledge, it becomes even more important to adaptively classify the synchronization dynamics to accurately characterize the intrinsic nature of seizure activities represented by the EEG. This study first measures the global synchronization by calculating Clustering Partition Mutual Information (MI) of all EEG data channels. A cross layer fully connected net is then designed to adaptively characterize the synchronization dynamics captured correlation matrices and automatically identify the seizure states of the EEG. Experiments are performed over the CHB-MIT scalp EEG dataset to evaluate the proposed approach. Seizure states can be identified with an accuracy, sensitivity and specificity of [98.19% #1; 0.24%], [98.27% #1; 0.51%], and [98.11% #1; 0.36%], respectively; the resulted performance is superior to those of most existing methods over the same dataset. The approach alleviates the need for strictly denoising and artifact removing based on the EEG prior knowledge that is mandatory for existing methods. Only one hyper-parameter need be set manually to avoid getting worse performance because of complex parameter setting.
Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2018-04-12 Cooperative journals: 《计算机应用研究》
Abstract： Aiming at security problems in smart grid, This paper propose a secure billing scheme based on privacy protection to interact and calculate a large amount of privacy data and improve the privacy data protection. Additive homomorphic encryption and mixed multiplication homomorphic encryption ensure the security of real-time power data in communication, data aggregation, electricity expenditure calculation and billing verification. Meanwhile, 爐he aggregation signature technology has reduced the overhead of data authentication process. The security analysis and performance analysis of the proposed scheme show that the scheme has good security and high performance.
Hits
Hits
Downloads
Comment