To address the challenges of traditional convolution in extracting neighborhood feature information accurately and integrating contextual information effectively in point cloud processing， a point cloud classification and segmentation method based on adaptive dynamic graph convolution and parameter-free attention was proposed. Firstly， the Adaptive Dynamic Graph Convolution module （ADGC） was used to learn feature information of different neighborhoods， generate the adaptive convolution kernels， and update the edge features， thereby extracting local neighborhood features of the point cloud accurately. Then， a residual structure was designed to learn spatial position information of the point cloud， so as to capture geometric structure between the point pairs accurately， and better retain and extract the detailed features. Finally， in order to better pay attention to and extract the local geometric features， the Parameter-Free Attention module （PFA） was combined with convolution operation to enhance connection among the neighbors and improve context-aware ability of the model. Experimental results show that compared to PointNet， the proposed method has significant advantages across various tasks. In specific， the proposed method has an increase of 4.6 percentage points in Overall Accuracy （OA） for classification tasks， an increase of 2.3 percentage points in mean Intersection over Union （mIoU） for part segmentation tasks， and an increase of 24.6 percentage points in mIoU for semantic segmentation tasks. It can be seen that the proposed method further improves the understanding and representation abilities of complex geometries， resulting in more accurate feature extraction and experimental performance in a variety of tasks.

Image data in actual business scenarios usually presents the characteristics of rich content and complex distortion performance， which is a great challenge to the generalization of objective Image Quality Assessment （IQA） algorithms. In order to solve this problem， a No-Reference IQA （NR-IQA） algorithm was proposed， which is mainly composed of three sub-networks： Feature Extraction Network （FEN）， Feature Fusion Network （FFN）， and Adaptive Prediction Network （APN）. Firstly， the global view， local patch， and saliency view of the sample were input into the FEN together， and the global distortion， local distortion， and saliency features were extracted by Swim Transformer. Then， the cascaded Transformer encoder was used to fuse the global distortion features and local distortion features， and the potential correlation patterns of the two were explored. Inspired by the human visual attention mechanism， the saliency features were used in the FFN to activate the attention module， so that the module was able to pay additional attention to the visual salient region， so as to improve the semantic parsing ability of the algorithm. Finally， the prediction score was calculated by the dynamically constructed MultiLayer Perceptron （MLP） regression network. Experimental results on main stream synthetic and real-world distortion datasets show that compared with the DSMix （Distortion-induced Sensitivity map-guided Mixed augmentation） algorithm， the proposed algorithm improves the Spearman Rank-order Correlation Coefficient （SRCC） by 4.3% on TID2013 dataset， and the Pearson Linear Correlation Coefficient （PLCC） by 1.4% on KonIQ dataset. The proposed algorithm also demonstrates excellent generalization ability and interpretability， which can deal with the complex distortion performance in business scenarios effectively， and can make adaptive prediction according to the individual characteristics of the sample.

To solve the problem that the traditional hand-eye calibration algorithms have large solution errors due to the noise interference in the processes of vision sensor calibration and robot kinematics solution， a robot hand-eye calibration algorithm based on Covariance Matrix Adaptation Evolutionary Strategy （CMAES） was proposed. Firstly， the mathematical tool Dual Quaternion （DQ） was used to establish the objective functions and geometric constraints for both rotation and translation， and the solution model was simplified. Then， the penalty function method was used to transform the constrained problem into an unconstrained optimization problem. Finally， CMAES algorithm was used to approximate the global optimal solution of hand-eye calibration rotation and translation equations. An experimental platform of robot and camera measurement was built， and the proposed algorithm was compared with two-step Tsai algorithm， the nonlinear optimization algorithm INRIA， and the DQ algorithm. Experimental results show that the solution error and variance of the proposed algorithm are smaller than those of traditional algorithms for both rotation and translation. Compared with Tsai algorithm， the proposed algorithm has the rotation accuracy improved by 4.58%， and the translation accuracy improved by 10.54%. It can be seen that the proposed algorithm has better solution accuracy and stability in the actual hand-eye calibration process with noise interference.

In order to cope with the current situations of low self-discipline， frequent violation events and difficult government supervision of enterprises in the internet environment， a Two-layer ensemble residual prediction Neural Network （TenrepNN） model was proposed to evaluate the self-discipline of enterprises. And by integrating the ideas of Stacking and Bagging ensemble learning， a new paradigm of integrated learning was designed， namely Adjusting. TenrepNN model has a two-layer structure. In the first layer， three base learners were used to predict the enterprise score preliminarily. In the second layer， the idea of residual correction was adopted， and a residual prediction neural network was proposed to predict the output deviation of each base learner. Finally， the final output was obtained by adding the deviations and the base learner scores together. On the enterprise self-discipline evaluation dataset， compared with the traditional neural network， the proposed model has the Root Mean Square Error （RMSE） reduced by 2.7%， and the classification accuracy in the self-discipline level reached 94.51%. Experimental results show that by integrating different base learners to reduce the variance and using residual prediction neural network to decrease the deviation explicitly， TenrepNN model can accurately evaluate enterprise self-discipline to achieve differentiated dynamic supervision.

Compared with 2D RGB-based images， 3D point clouds retain the real and rich geometric information of objects in space to deal with vision challenge with scale variation in the single object tracking problem. However， the precision of 3D object tracking is affected by the loss of information brought by the sparsity of point cloud data and the deformation caused by the object position changing. To solve the above two problems， a proposal-based aggregation network composed of three modules was proposed in an end-to-end learning pattern. In this network， the 3D bounding box was determined by locating object center in the best proposal to realize the single object tracking in 3D point cloud. Firstly， the point cloud data of both templates and search areas was transferred into bird’s-eye view pseudo images. In the first module， the feature information was enriched through spatial and cross-channel attention mechanisms. Then， in the second module， the best proposal was given by the anchor-based deep cross-correlation Siamese region proposal subnetwork. Finally， in the third module， the object features were extracted through region of interest pooling operation by the best proposal at first， and then， the object and template features were aggregated， the sparse modulated deformable convolution layer was used to deal with the problems of point cloud sparsity and deformation， and the final 3D bounding box was determined. Experimental results of the comparison between the proposed method and the state-of-the-art 3D point cloud single object tracking methods on KITTI dataset show that： in comprehensive experiment of car， the proposed method has improved 1.7 percentage points on success rate and 0.2 percentage points on precision in real scenes； in multi-category extensive experiment of car， van， cyclist and pedestrian， the proposed method has improved the average success rate by 0.8 percentage points， and the average precision by 2.8 percentage points， indicating that the proposed method can solve the single object tracking problem in 3D point cloud and make the 3D object tracking results more accurate.

Recently, Extreme Learning Machine (ELM) has been a promising tool in solving a wide range of classification and regression problems. However, the generalization performance of ELM will be decreased when there exits redundant hidden neurons. To solve this problem, this paper introduced a new regularizer that was the Frobenius norm of mapping matrix from hidden space to a new latent feature space. Furthermore, an alternating optimization strategy was adopted to learn the above regularization problem and the latent feature space. The proposed algorithm was tested empirically on the classical UCI data set as well as a load identification engineering data set. The experimental results show that the proposed algorithm obviously outperforms the classical ELM in terms of predictive precision and numerical stability, and needs much less computational cost than the present ELM model selection algorithm.

To solve the problem that the traditional Multi-Input-Multi-Output (MIMO) Support Vector Machine (SVM) generally ignore the dependency among all outputs, a new MIMO SVM algorithm based on principal curve was proposed in this paper. Following the assumption that the model parameters of all outputs locate on a manifold, this paper firstly constructed a manifold regularization based on the Multi-dimensional Support Vector Regression (M-SVR), where the regularizer was the squared distance from the output parameters to the principal curve through the middle of all parameters' set. Secondly, considering the non-convexity of this regularization, this paper introduced an alternative optimization method to calculate the model parameters and principal curve in turn until convergence. The experiments on simulated data and real-life dynamic load identification data were conducted, and the results show that the proposed algorithm performs better than M-SVR and SVM based separate modeling method in terms of prediction precision and numerical stability.