In the last decade, legal case search has become an important part of a legal practitioner's work. During legal case search, search engines retrieval a number of relevant cases from huge amounts of data and serve them to users. However, it is uncertain whether these cases are gender-biased and whether such bias has impact on user perceptions. We designed a new user experiment framework to simulate the judges' reading of relevant cases. 72 participants with backgrounds in legal affairs invited to conduct the experiment. Participants were asked to simulate the role of the judge in conducting a legal case search on 3 assigned cases and determine the sentences of the defendants in these cases. Gender of the defendants in both the task and relevant cases was edited to statistically measure the effect of gender bias in the legal case search results on participants' perceptions. The results showed that gender bias in the legal case search results did not have a significant effect on judges' perceptions.
Complex dynamic network data,such as social networks,protein interaction networks,and infectious disease transmission networks,are prevalent in the real world,consisting of numerous nodes and edges.Effective mining and utili-zation of such data for accurate prediction have become a key task.Dynamic graph neural network link prediction is an important branch of deep learning research,which aims to analyze the intrinsic laws of network evolution over time and predict potential future linkages,providing valuable information and basis for decision-making in various fields.This paper first reviews the development of dynamic graph neural networks,then introduces the modeling methods and training processes of dynamic graphs.Based on this,the paper categorizes dynamic graph neural network link prediction models into two main types according to the granularity of time:discrete dynamic graph models and continuous dynamic graph models,and provides an overview of the modeling methods used by current mainstream models in each category.In addi-tion,it also introduces commonly used data sets,evaluation indicators and some application scenarios in dynamic graph link prediction research.Finally,the future development trends in this field are discussed prospectively.
Accurate and reliable short-term forecasting of passenger flows can ensure the operation of urban rail trans-port.Considering the differences in the timing characteristics of passenger flows at different stations,a deep learning method for predicting passenger flows at urban stations is developed based on station classification.Firstly,stations are classified by dynamic time warping and the K-means algorithm,and the timing characteristics of the passenger flow of various stations are analyzed.Secondly,the complete ensemble empirical mode decomposition with adaptive noise is used to decompose passenger flow data of various stations to reduce the effects of data noise.Finally,a deep learning pre-diction method integrating long short-term memory and Transformer model is proposed to predict the passenger flow of different types of stations.The method is verified by using the passenger flow data of Xi'an Metro.The results show that the stations can be classified into four types according to the timing characteristics of passenger flow data on working days and non-working days:occupation-residential balance type,business office type,leisure and entertainment type,and dense residential type.Compared with the other three single models and three combined models,mean absolute error of passenger flow prediction results of the proposed method in different types of stations is reduced by 16.36%～51.02%,root mean square error is reduced by 10.35%～50.76%,and mean absolute percentage error is reduced by 14.71%～48.62%.Compared with the other six models,the prediction results of the station passenger flow data based on the statis-tics of different time intervals of 15 min,30 min,45 min and 60 min,the three indicators are respectively reduced by 12.63%～51.02%,8.08%～49.12%and 6.83%～47.26%.
In the application of driverless technology, current traffic sign recognition methods are susceptible to the influence of ambient light interference, target size changes and complex backgrounds, resulting in reduced recognition accuracy. To address these challenges, this study introduces an optimisation algorithm called ETSR-YOLO, which is based on the YOLOv5s algorithm. First, this study improves the path aggregation network (PANet) of YOLOv5s to enhance multi-scale feature fusion by generating an additional high-resolution feature layer to improve the recognition of YOLOv5s for small-sized objects. Second, the study introduces two improved C3 modules that aim to suppress background noise interference and enhance the feature extraction capabilities of the network. Finally, the study uses the Wise-IoU (WIoU) function in the post-processing stage to improve the learning ability and robustness of the algorithm to different samples. The experimental results show that ETSR-YOLO improves mAP@0.5 by 6.6% on the Tsinghua-Tencent 100K (TT100K) dataset and by 1.9% on the CSUST Chinese Traffic Sign Detection Benchmark 2021 (CCTSDB2021) dataset. In the experiments conducted on the embedded computing platform, ETSR-YOLO demonstrates a short average inference time, thereby affirming its capability to deliver dependable traffic sign detection for intelligent vehicles operating in real-world traffic scenes. The source code and test results of the models used in this study are available at https://github.com/cbrook16/ETSR-YOLO.
Remote sensing scene classification is a hot research topic in computer vision and it is of great significance to the semantic understanding of remote sensing images. At present, remote sensing scene classification methods based on deep learning occupy a dominant position in this field. However, it suffers from the lack of samples and poor model generalization ability in actual application scenarios. Therefore, this paper proposes a few-shot remote scene classification method based on attention mechanism, and designs a structure of dual-branches similarity measurement. This method is based on the meta-learning training strategy to divide the dataset into tasks. At the meantime, the input images are divided into blocks to preserve the feature distribution in the remote sensing image. Then the lightweight attention module is introduced into the feature extraction network to reduce the risk of overfitting and ensure the acquisition of discriminative features. Finally we add a dual-branches similarity measurement module on the basis of Earth Mover’s Distance to improve the discriminative ability of the classifier. The results show that compared with the classic small-sample learning method, the few-shot remote scene classification method proposed in this paper can significantly improve the classification performance.
