Traditional data augmentation techniques， such as synonym substitution， random insertion， and random deletion， may change the original semantics of text and even result in the loss of critical information. Moreover， data in text classification tasks typically have both textual and label parts. However， traditional data augmentation methods only focus on the textual part. To address these issues， a Label Confusion incorporated Data Augmentation （LCDA） technique was proposed for providing a comprehensive enhancement of data from both textual and label aspects. In terms of text， by enhancing the text through random insertion and replacement of punctuation marks and completing end-of-sentence punctuation marks， textual diversity was increased with all textual information and sequence preserved. In terms of labels， simulated label distribution was generated using a label confusion approach， and used to replace the traditional one-hot label distribution， so as to better reflect the relationships among instances and labels as well as between labels. In experiments conducted on few-shot datasets constructed from THUCNews （TsingHua University Chinese News） and Toutiao Chinese news datasets， the proposed technique was combined with TextCNN， TextRNN， BERT （Bidirectional Encoder Representations from Transformers）， and RoBERTa-CNN （Robustly optimized BERT approach Convolutional Neural Network） text classification models. The experimental results indicate that compared to those before enhancement， all models demonstrate significant performance improvements. Specifically， on 50-THU， a dataset constructed on THUCNews dataset， the accuracies of four models combing LCDA technique are improved by 1.19， 6.87， 3.21， and 2.89 percentage points， respectively， compared to those before enhancement， and by 0.78， 7.62， 1.75， and 1.28 percentage points， respectively， compared to those of the four models combining softEDA （Easy Data Augmentation with soft labels） method. By both textual and label processing results， model accuracy is enhanced by LCDA technique significantly， particularly in application scenarios characterized by limited data availability.
With the development of lightweight networks， human pose estimation tasks can be performed on devices with limited computational resources. However， improving accuracy has become more challenging. These challenges mainly led by the contradiction between network complexity and computational resources， resulting in the sacrifice of representation capabilities when simplifying the model. To address these issues， a Decoupled attention and Ghost convolution based Lightweight human pose estimation Network （DGLNet） was proposed. Specifically， in DGLNet， with Small High-Resolution Network （Small HRNet） model as basic architecture， by introducing a decoupled attention mechanism， DFDbottleneck module was constructed. The basic modules were redesigned with shuffleblock structure， in which computationally-intensive point convolutions were replaced with lightweight ghost convolutions， and the decoupled attention mechanism was utilized to enhance module performance， leading to the creation of DGBblock module. Additionally， the original transition layer modules were replaced with redesigned depthwise separable convolution modules that incorporated ghost convolution and decoupled attention， resulting in the construction of GSCtransition module. This modification further reduced computational complexity while enhancing feature interaction and performance. Experimental results on COCO validation set show that DGLNet outperforms the state-of-the-art Lite-High-Resolution Network （Lite-HRNet） model， achieving the maximum accuracy of 71.9% without increasing computational complexity or the number of parameters. Compared to common lightweight pose estimation networks such as MobileNetV2 and ShuffleNetV2， DGLNet achieves the precision improvement of 4.6 and 8.3 percentage points respectively， while only utilizing 21.2% and 25.0% of their computational resources. Furthermore， under the AP50 evaluation criterion， DGLNet surpasses the large High-Resolution Network （HRNet） while having significantly less computational and parameters.
As a kind of side information， Knowledge Graph （KG） can effectively improve the recommendation quality of recommendation models， but the existing knowledge-awareness recommendation methods based on Graph Neural Network （GNN） suffer from unbalanced utilization of node information. To address the above problem， a new recommendation method based on Knowledge?awareness and Cross-level Contrastive Learning （KCCL） was proposed. To alleviate the problem of unbalanced node information utilization caused by the sparse interaction data and noisy knowledge graph that deviate from the true representation of inter-node dependencies during information aggregation， a contrastive learning paradigm was introduced into knowledge-awareness recommendation model of GNN. Firstly， the user-item interaction graph and the item knowledge graph were integrated into a heterogeneous graph， and the node representations of users and items were realized by a GNN based on the graph attention mechanism. Secondly， consistent noise was added to the information propagation aggregation layer for data augmentation to obtain node representations of different levels， and the obtained outermost node representation was compared with the innermost node representation for cross-level contrastive learning. Finally， the supervised recommendation task and the contrastive learning assistance task were jointly optimized to obtain the final representation of each node. Experimental results on DBbook2014 and MovieLens-1m datasets show that compared to the second prior contrastive method， the Recall@10 of KCCL is improved by 3.66% and 0.66%， respectively， and the NDCG@10 is improved by 3.57% and 3.29%， respectively， which verifies the effectiveness of KCCL.
Due to the ambiguity of text and the lack of location information in training data， current state-of-the-art diffusion model cannot accurately control the locations of generated objects in the image under the condition of text prompts. To address this issue， a spatial condition of the object’s location range was introduced， and an attention-guided method was proposed based on the strong correlation between the cross-attention map in U-Net and the image spatial layout to control the generation of the attention map， thus controlling the locations of the generated objects. Specifically， based on the Stable Diffusion （SD） model， in the early stage of the generation of the cross-attention map in the U-Net layer， a loss was introduced to stimulate high attention values in the corresponding location range， and reduce the average attention value outside the range. The noise vector in the latent space was optimized step by step in each denoising step to control the generation of the attention map. Experimental results show that the proposed method can significantly control the locations of one or more objects in the generated image， and when generating multiple objects， it can reduce the phenomenon of object omission， redundant object generation， and object fusion.
The flood of fake job advertisements will not only damage the legitimate rights and interests of job seekers but also disrupt the normal employment order， which results in a poor user experience for job seekers. To effectively detect fake job advertisements， an SSC （Semi-Supervised fake job advertisements detection model based on Consistency training） was proposed. Firstly， the consistency regularization term was applied on all the data to improve the performance of the model. Then， supervised loss and unsupervised loss were integrated through joint training to obtain the semi-supervised loss. Finally， the semi-supervised loss was used to optimize the model. Experimental results on two real datasets EMSCAD （EMployment SCam Aegean Dataset） and IMDB （Internet Movie DataBase） show that SSC achieves the best detection performance when the labeled data are only 20， and the accuracy is increased by 2.2 and 2.8 percentage points compared with the existing advanced semi-supervised learning model UDA （Unsupervised Data Augmentation）， and is increased by 3.4 and 11.7 percentage points compared with the deep learning model BERT （Bidirectional Encoder Representations from Transformers）. At the same time， SSC has good scalability.
Attribute reduction is a hot research topic in rough set theory. Most of the algorithms of attribute reduction for continuous data are based on dominance relations or neighborhood relations. However， continuous datasets do not necessarily have dominance relations in attributes. And the attribute reduction algorithms based on neighborhood relations can adjust the granulation degree through neighborhood radius， but it is difficult to unify the radii due to the different dimensions of attributes and the continuous values of radius parameters， resulting in high computational cost of the whole parameter granulation process. To solve this problem， a multi-granularity attribute reduction strategy based on cluster granulation was proposed. Firstly， the similar samples were classified by the clustering method， and the concepts of approximate set， relative positive region and positive region reduction based on clustering were proposed. Secondly， according to JS （Jensen-Shannon） divergence theory， the difference of data distribution of each attribute among clusters was measured， and representative features were selected to distinguish different clusters. Finally， an attribute reduction algorithm was designed using a discernibility matrix. In the proposed algorithm， the attributes were not required to have ordered relations. Different from neighborhood radius， the clustering parameter was discrete， and the dataset was able to be divided into different granulation degrees by adjusting this parameter. Experimental results on UCI and Kent Ridge datasets show that this attribute reduction algorithm can directly deal with continuous data. At the same time， by using this algorithm， the redundant features in the datasets can be removed while maintaining or even improving the classification accuracy by discrete adjustment of the parameters in a small range.
Meta-learning is the learning process of applying machine learning methods （meta-algorithms） to seek the mapping between features of a problem （meta-features） and relative performance measures of the algorithm， thereby forming the learning process of meta-knowledge. How to construct and extract meta-features is an important research content. Concerning the problem that most of meta-features used in the existing related researches are statistical features of data， uncertainty modeling was proposed and the impact of uncertainty on learning system was studied. Based on inconsistency of data， complexity of boundary， uncertainty of model output， linear capability to be classified， degree of attribute overlap， and uncertainty of feature space， six kinds of uncertainty meta-features were established for data or models. At the same time，the uncertainty size of the learning problem itself was measured from different perspectives， and specific definitions were given. The correlations between these meta-features were analyzed on artificial datasets and real datasets of a large number of classification problems， and multiple classification algorithms such as K-Nearest Neighbor （KNN） were used to conduct a preliminary analysis of the correlation between meta-features and test accuracy. Results show that the average degree of correlation is about 0.8， indicating that these meta-features have a significant impact on learning performance.
Concerning the problem that existing blind road recognition method has low recognition rate, simplistic handling, and is easily influenced by light, or shadow, an improved blind road recognition method was proposed. According to the color and texture features of blind road, the algorithm used two segmentation methods respectively including color histogram feature threshold segmentation combined with improved region growing segmentation and fuzzy C-means clustering segmentation for gray level co-occurrence matrix feature. And combined with Canny edge detection and Hough transform algorithm, the proposed algorithm made the blind area separated from the pedestrian area and determines the migration direction for the blind. The experimental results show that the proposed algorithm can segment several kinds of blind road more accurately, detect the boundary and direction of blind road and solve the light and shadow problem partly. It can choose the fastest and the most effective segmentation method adoptively, and can be used in a variety of devices, such as electronic guide ones.
