首页|基于胶囊网络的多标签罪名预测

基于胶囊网络的多标签罪名预测

扫码查看
原文链接
  • NETL
  • NSTL
  • 万方数据
  • 维普
针对罪名预测任务中存在的数据不平衡,罪名预测效果不理想的问题,提出了基于胶囊网络的多标签罪名预测模型.使用双向门控循环神经网络与卷积神经网络的并行混合模型提取胶囊网络的初始特征,即提取数据的时序特征和语义特征,提高胶囊网络提取初始特征的能力,然后使用动态路由进行迭代运算提取深层次的空间的信息;在损失函数中引入权值信息,解决因数据不平衡而导致的低频数据训练不足的问题.仿真结果表明,相比其它经典深度学习模型,上述模型有较好的罪名预测效果,能够解决数据不平衡与罪名预测效果不理想的问题.
Prediction of Multi-Label Charges Based on Capsule Network
In view of the imbalance of data and the unsatisfactory effect of charge prediction in the task of charge prediction,a multi-label crime prediction model based on capsule network was proposed.The parallel hybrid model of bi-directional gated recurrent neural network and convolutional neural network is used to extract the initial features of the capsule network,that is,to extract the temporal and semantic features of the data,so as to improve the ability of the capsule network to extract the initial features,and then the dynamic route is used for iterative operations to extract the deep-seated spatial information.The weight information was introduced into the loss function to solve the problem of insufficient training of low-frequency data caused by data imbalance.The simulation results show that compared with other classical deep learning models,this model has better charge prediction effect,and can solve the problems of data imbalance and unsatisfactory charge prediction effect.

Accusation predictionData imbalanceCapsule networkWeight informationLow frequency data

王之原、张琛、胡叮叮

展开 >

甘肃政法大学网络空间安全学院,甘肃 兰州 730070

罪名预测 数据不平衡 胶囊网络 权重信息 低频数据

2022 省级产业支撑项目甘肃政法大学2020年校级教改项目甘肃省教育厅高等学校科研项目甘肃省软科学项目甘肃省 2020年省级虚拟仿真一流课程甘肃省教育厅青年博士基金项目

2022CYZC-57GZJG2020-B062015A-11420CX4ZA074GZYL2020-182022QB-132

2024

计算机仿真
中国航天科工集团公司第十七研究所

计算机仿真

CSTPCD
影响因子：0.518
ISSN：1006-9348
年,卷(期)：2024.41(1)
  • 19