Aiming at the fact that the existing video person re-identification methods cannot effectively extract the spatiotemporal information between consecutive frames of the video, a person re-identification network based on non-local attention and multi-feature fusion was proposed to extract global and local representation features and time series information. Firstly, the non-local attention module was embedded to extract global features. Then, the multi-feature fusion was realized by extracting the low-level and middle-level features as well as the local features, so as to obtain the salient features of the person. Finally, the similarity measurement and sorting were performed to the person features in order to calculate the accuracy of video person re-identification. The proposed model has significantly improved performance compared to the existing Multi-scale 3D Convolution (M3D) and Learned Clip Similarity Aggregation (LCSA) models with the mean Average Precision (mAP) reached 81.4% and 93.4% respectively and the Rank-1 reached 88.7% and 95.3% respectively on the large datasets MARS and DukeMTMC-VideoReID. At the same time, the proposed model has the Rank-1 reached 94.8% on the small dataset PRID2011.

In the real world, multi-label text has a wider application scenario than single-label text. At the same time, due to its huge output space, it brings a lot of challenges to the classification task. The multi-label text classification problem was regarded as label sequence generation problem, and the Sequence Generation Model (SGM) was applied to the multi-label text classification field. Aiming at the problems such as that the sequence structure of the model is easy to produce the cumulative error, an SGM based on Dynamic Routing (DR-SGM) was proposed. The model was based on Encoder-Decoder mode. In the Encoder layer, Bi-directional Long Short-Term Memory (Bi-LSTM) neural network+Attention was used to encode the semantic information. In the Decoder layer, a decoder structure with the dynamic routing aggregation layer was designed which reduces the influence of the cumulative error added behind the hidden layer. At the same time, the part-part and part-glob position information in the text was captured by dynamic routing. And by optimizing the dynamic routing algorithm, the semantic clustering effect was further improved. DR-SGM was applied to the classification of multi-label texts. The experimental results show that DR-SGM improves multi-label text classification results on the RCV1-V2, AAPD and Slashdot datasets.

A state-space model was established for the short-term traffic flow prediction problem under complex road conditions, which is based on macroscopic traffic flow forecasting. In order to solve the problem of parameter optimization on the dynamic traffic forecast model, a method to improve the performance of Unscented Kalman Filter (UKF) with orthogonal adaptive Differential Evolution (DE) was proposed. The orthogonal method maximized the diversity of the initial population in DE algorithm. The crossover operator in DE was optimized by the orthogonal method and the technology of quantification to balance the exploitation and exploration, which was more beneficial to find the model parameters of UKF. The experimental results show that, with respect to use random distribution to initialize the parameters, or set model parameters based on the experience, the use of orthogonal design method for initialization strategy, mutation operator and adaptive control strategy of parameters in differential evolution algorithm can effectively save computing resources, improve forecasting performance and accuracy, and provide better robustness.

When the Synthetic Minority Over-sampling Technique (SMOTE) is used in imbalance dataset classification, it sets the same sampling rate for all the samples of minority class in the process of synthetising new samples, which has blindness. To overcome this problem, a Genetic Algorithm (GA) improved SMOTE algorithm, namely GASMOTE (Genetic Algorithm Improved Synthetic Minority Over-sampling Technique) was proposed. At the beginning, GASMOTE set different sampling rates for different minority class samples. One combination of the sampling rates corresponded to one individual in the population. And then, the selection, crossover and mutation operators of GA were iteratively applied on the population to get the best combination of sampling rates when the stopping criteria were met. At last, the best combination of sampling rates was used in SMOTE to synthetise new samples. The experimental results on ten typical imbalance datasets show that, compared with SMOTE algorithm, GASMOTE can increase 5.9 percentage on F-measure value and 1.6 percentage on G-mean value, and compared with Borderline-SMOTE algorithm, GASMOTE can increase 3.7 percentage on F-measure value and 2.3 percentage on G-mean value. GASMOTE can be used as a new over-sampling technique to deal with imbalance dataset classification problem.

A new autonomous vehicle navigation model was proposed based on multi-sensor system for vehicle navigation and Global Positioning System (GPS) under complex road conditions. And the Unscented Kalman Filter (UKF) was used to overcome some security issues due to the sudden error produced by the Kalman filters with extensions, which belonged to Sigma point based sensor fusion algorithm. It is more suitable than the Kalman filters with extensions that the UKF can calculate the evaluation satisfied the requirement in vehicle navigation. Comparison experiments with the Kalman filter based on polynomial expansion were given in terms of estimation accuracy and computational speed. The experimental results show that the Sigma-point Kalman filter is a reliable and computationally efficient approach to state estimation-based control. Moreover, it is faster to evaluate the motion state of the vehicle according to the current direction situations and the feedback information of vehicle sensor, and can calculate the control input of vehicle adaptively in real time.