A Hybrid Wavelet Transform Based Short-Term Wind Speed Forecasting Approach by Jujie Wang

“…In the second step, a two-hidden-layer neural network (TNN) is used to predict both approximated scale and detailed scales, respectively. …”
Get full text

Differential privacy budget optimization based on deep learning in IoT by Dan LUO, Ruzhi XU, Zhitao GUAN

“…In order to effectively process the massive data brought by the large-scale application of the internet of things (IoT), deep learning is widely used in IoT environment.However, in the training process of deep learning, there are security threats such as reasoning attacks and model reverse attacks, which can lead to the leakage of the original data input to the model.Applying differential privacy to protect the training process parameters of the deep model is an effective way to solve this problem.A differential privacy budget optimization method was proposed based on deep learning in IoT, which adaptively allocates different budgets according to the iterative change of parameters.In order to avoid the excessive noise, a regularization term was introduced to constrain the disturbance term.Preventing the neural network from over fitting also helps to learn the salient features of the model.Experiments show that this method can effectively enhance the generalization ability of the model.As the number of iterations increases, the accuracy of the model trained after adding noise is almost the same as that obtained by training using the original data, which not only achieves privacy protection, but also guarantees the availability, which means balance the privacy and availability.…”
Get full text

Automatic Recognition and Detection System Based on Machine Vision by Lu Wang, Lihong Cui

“…This paper combines machine vision with motion control theory and uses pulse coupled neural network (PCNN) edge detection and recognition algorithms to preliminarily design a set of machine vision automatic recognition and detection systems and carry out detection and recognition experiments on small parts such as relay covers. …”
Get full text

Research on unsupervised domain adaptive bearing fault diagnosis method by WU ShengKai, SHAO Xing, WANG CuiXiang, GAO Jun

“…Firstly, the bearing fault sample data was preprocessed by fast Fourier transform and the features of bearing faults samples were extracted using convolutional neural network. Then, the feature distributions output of the source domain and the target domain were converged by the method of reversing labels in the generative adversarial network. …”
Get full text

Network threat situation assessment based on unsupervised multi-source data feature analysis by Hongyu YANG, Fengyan WANG

“…Aiming at the limitations of supervised neural network in the network threat testing task relying on data category tagging,a network threat situation evaluation method based on unsupervised multi-source data feature analysis was proposed.Firstly,a variant auto encoder-generative adversarial network (V-G) for security threat assessment was designed.The training data set containing only normal network traffic was input to the network collection layer of V-G to perform the model training,and the reconstruction error of the network output of each layer was calculated.Then,the reconstruction error learning was performed by the three-layer variation automatic encoder of the output layer,and the training abnormal threshold was obtained.The packet threat was tested by using the test data set containing the abnormal network traffic,and the probability of occurrence of the threat of each group of tests was counted.Finally,the severity of the network security threat was determined according to the probability of threat occurrence,and the threat situation value was calculated according to the threat impact to obtain the network threat situation.The simulation results show that the proposed method has strong characterization ability for network threats,and can effectively and intuitively evaluate the overall situation of network threat.…”
Get full text

Network intrusion detection method based on VAE-CWGAN and fusion of statistical importance of feature by Taotao LIU, Yu FU, Kun WANG, Xueyuan DUAN

“…Considering the problems of traditional intrusion detection methods limited by the class imbalance of datasets and the poor representation of selected features, a detection method based on VAE-CWGAN and fusion of statistical importance of features was proposed.Firstly, data preprocessing was conducted to enhance data quality.Secondly, a VAE-CWGAN model was constructed to generate new samples, addressing the problem of imbalanced datasets, ensuring that the classification model no longer biased towards the majority class.Next, standard deviation, difference of median and mean were used to rank the features and fusion their statistical importance for feature selection, aiming to obtain more representative features, which made the model can better learn data information.Finally, the mixed data set after feature selection was classified through a one-dimensional convolutional neural network.Experimental results show that the proposed method demonstrates good performance advantages on three datasets, namely NSL-KDD, UNSW-NB15, and CIC-IDS-2017.The accuracy rates are 98.95%, 96.24%, and 99.92%, respectively, effectively improving the performance of intrusion detection.…”
Get full text

Statistical mechanics and machine learning of the α-Rényi ensemble by Andrew Jreissaty, Juan Carrasquilla

“…We conclude by performing a variational minimization of the α-Rényi free energy using a recurrent neural network (RNN) Ansatz where we find that the RNN performs well in two dimensions when compared to the Monte Carlo simulations. …”
Get full text

Few Samples of SAR Automatic Target Recognition Based on Enhanced-Shape CNN by Mengmeng Huang, Fang Liu, Xianfa Meng

“…Some researchers point out that existing convolutional neural network (CNN) paid more attention to texture information, which is often not as good as shape information. …”
Get full text

Speaker verification method based on cross-domain attentive feature fusion by Zhen YANG, Tianlang WANG, Haiyan GUO, Tingting WANG

“…Aiming at the problem that the lack of structure information among speech signal sample in the front-end acoustic features of speaker verification system, a speaker verification method based on cross-domain attentive feature fusion was proposed.Firstly, a feature extraction method based on the graph signal processing (GSP) was proposed to extract the structural information of speech signals, each sample point in a speech signal frame was regarded as a graph node to construct the speech graph signal and the graph frequency information of the speech signal was extracted through the graph Fourier transform and filter banks.Then, an attentive feature fusion network with the residual neural network and the squeeze-and- excitation block was proposed to fuse the features in the traditional time-frequency domain and those in the graph frequency domain to promote the speaker verification system performance.Finally, the experiment was carried out on the VoxCeleb, SITW, and CN-Celeb datasets.The experimental results show that the proposed method performs better than the baseline ECAPA-TDNN model in terms of equal error rate (EER) and minimum detection cost function (min-DCF).…”
Get full text

Forwarding efficiency aware traffic scheduling algorithm based on deep reinforcement learning by Zongxuan SHA, Ru HUO, Chuang SUN, Shuo WANG, Tao HUANG

“…The software defined network separates the control plane from the data plane to achieve flexible traffic scheduling, which can use network resources more efficiently.However, with the increase of the number of flow entries, load rate, the number of connected hosts, and other factors, the forwarding efficiency of the SDN switch will be reduced, which will affect the end-to-end transmission delay.To solve the above problems, the forwarding efficiency aware traffic scheduling algorithm based on deep reinforcement learning was proposed.First, the switch state was integrated into the perception model, and the mapping relationship between switch state information and forwarding efficiency was established based on neural network.Then, combined with network state and traffic information, traffic scheduling policy was generated through deep reinforcement learning.Finally, the expert samples generated by the shortest path and load balance algorithms could guide the model training, which enabled the model to learn knowledge from expert samples to improve performance and accelerated the training process.The experimental results show that the proposed algorithm not only reduces the average end-to-end transmission delay by 15.31%, but also ensures the overall load balance of the network, compared with other algorithms.…”
Get full text

Neural processing of naturalistic audiovisual events in space and time by Yu Hu, Yalda Mohsenzadeh

“…Comparing neural representations to a two-branch deep neural network model highlighted the necessity of early cross-modal connections to build a biologically plausible model of audiovisual perception. …”
Get full text

Data Preprocessing Method and Fault Diagnosis Based on Evaluation Function of Information Contribution Degree by Siyu Ji, Chenglin Wen

“…Neural network is a data-driven algorithm; the process established by the network model requires a large amount of training data, resulting in a significant amount of time spent in parameter training of the model. …”
Get full text

Digital watermarking method based on context word prediction and window compression coding by Lingyun XIANG, Minghao HUANG, Chenling ZHANG, Chunfang YANG

“…To address the problems of limited number of substitutable words and low watermark extraction efficiency in the existing natural language digital watermarking methods, a creative method based on context word prediction and window compression coding was proposed.Firstly, the contextual semantic features of each word in the original text were automatically learned through a neural network language model, and then the candidate word set for each word was predicted, thus the number of substitutable words that could be utilized for carrying watermark information was expanded.Meanwhile, considering the difference of the semantic impact caused by the substitutions of candidate words at different positions, the watermark information was embedded into each window containing several words, and the selection of candidate words for watermark embedding was optimized by the similarity between sentences before and after performing word substitutions.Finally, a semantic-independent window compression coding method was proposed, which encoded each window as appointed watermark information in terms of the character information of words contained in the window.So that during watermark extraction, the dependence on the original context at the position of word substitution was eliminated.The experimental results show that the proposed method greatly improves the watermark extraction efficiency with high embedding capacity and text quality.…”
Get full text

Design and realization of compressor data abnormality safety monitoring and inducement traceability expert system. by Yuan Wang, Shaolin Hu

“…The results show that this method effectively overcomes the problems of false alarms and missed alarms based on fixed threshold alarm methods, and achieves 100% classification of two types of faults: non starting of the drive machine and low oil pressure by constructing a PCA (Principal Component Analysis)-SPE (Square Prediction Error)-CNN (Convolutional Neural Network) classifier. Combined with dynamic knowledge graph and NLP (Natural Language Processing) inference, it achieves good diagnostic results.…”
Get full text

A Modified Fully Convolutional Network for Crack Damage Identification Compared with Conventional Methods by Meng Meng, Kun Zhu, Keqin Chen, Hang Qu

“…With the development of artificial intelligence especially the combination of deep learning and computer vision, greater advantages have been brought to the concrete crack detection based on convolutional neural network (CNN) over the traditional methods. However, these machine learning (ML) methods still have some defects, such as it being inaccurate or not strong, having poor generalization ability, or the accuracy still needs to be improved, and the running speed is slow. …”
Get full text

Metric-based learning approach to botnet detection with small samples by Honggang LIN, Junjing ZHU, Lin CHEN

“…Botnets pose a great threat to the Internet, and early detection is crucial for maintaining cybersecurity.However, in the early stages of botnet discovery, obtaining a small number of labeled samples restricts the training of current detection models based on deep learning, leading to poor detection results.To address this issue, a botnet detection method called BT-RN, based on metric learning, was proposed for small sample backgrounds.The task-based meta-learning training strategy was used to optimize the model.The verification set was introduced into the task and the similarity between the verification sample and the training sample feature representation was measured to quickly accumulate experience, thereby reducing the model’s dependence on the labeled sample space.The feature-level attention mechanism was introduced.By calculating the attention coefficients of each dimension in the feature, the feature representation was re-integrated and the importance attention was assigned to optimize the feature representation, thereby reducing the feature sparseness of the deep neural network in small samples.The residual network design pattern was introduced, and the skip link was used to avoid the risk of model degradation and gradient disappearance caused by the deeper network after increasing the feature-level attention mechanism module.…”
Get full text

Machine learning-based analyzing earthquake-induced slope displacement. by Jiyu Wang, Niaz Muhammad Shahani, Xigui Zheng, Jiang Hongwei, Xin Wei

“…This study evaluates the capabilities of various machine learning models, including artificial neural network (ANN), support vector machine (SVM), random forest (RF), and extreme gradient boosting (XGBoost) in analyzing earthquake-induced slope displacement. …”
Get full text

Personalized Travel Route Recommendation Model of Intelligent Service Robot Using Deep Learning in Big Data Environment by Xiang Huang

“…Firstly, by crawling the relevant website data, obtain the basic information data and comment the text data of tourism service items, as well as the basic information data, and comment the text data of users and preprocess them, such as data cleaning. Then, a neural network model based on the self-attention mechanism is proposed, in which the data features are obtained by the Gaussian kernel function and node2vec model, and the self-attention mechanism is used to capture the long-term and short-term preferences of users. …”
Get full text

Evaluating Machine Learning Models for Prostate Cancer Classification Using Gene Expression Profiles from DNA Microarrays by Haddou Bouazza Sara, Haddou Bouazza Jihad

“…These methods were combined with classifiers such as K Nearest Neighbor (KNN), Support Vector Machine (SVM), Linear Discriminant Analysis (LDA), Decision Tree Classifier (DTC), Naïve Bayes (NB), and Artificial Neural Network (ANN). Our results demonstrated that the best combination was the Signal to Noise Ratio with Linear Discriminant Analysis, achieving a classification accuracy of 95% using only six genes. …”
Get full text

Dilated SE-DenseNet for brain tumor MRI classification by Yuannong Mao, Jiwook Kim, Lena Podina, Mohammad Kohandel

“…Abstract In the field of medical imaging, particularly MRI-based brain tumor classification, we propose an advanced convolutional neural network (CNN) leveraging the DenseNet-121 architecture, enhanced with dilated convolutional layers and Squeeze-and-Excitation (SE) networks’ attention mechanisms. …”
Get full text