基于卷积神经网络和残差结构单元的合同数据识别提取.docx

基于卷积神经网络和残差结构单元的合同数据识别提取

目录

1.内容简述................................................2

1.1研究背景.............................................2

1.2研究意义.............................................3

1.3文献综述.............................................4

2.相关技术介绍............................................6

2.1卷积神经网络.........................................7

2.1.1CNN基本结构......................................8

2.1.2CNN工作原理......................................9

2.2残差结构单元........................................10

2.2.1ResNet结构特点..................................11

2.2.2ResNet在图像识别中的应用........................12

3.合同数据识别提取方法...................................14

3.1数据预处理..........................................15

3.1.1数据清洗........................................16

3.1.2数据增强........................................17

3.2网络结构设计........................................17

3.2.1卷积层设计......................................18

3.2.2残差单元设计....................................20

3.3损失函数与优化器....................................21

3.3.1损失函数........................................22

3.3.2优化器选择......................................22

4.实验与结果分析.........................................23

4.1实验环境............................................25

4.2数据集介绍..........................................25

4.3实验步骤............................................26

4.4结果分析............................................27

4.4.1模型性能评估....................................28

4.4.2消融实验........................................29

5.应用案例分析...........................................31

5.1案例一..............................................32

5.2案例二..............................................34

5.3案例三..............................................36

6.结论与展望.............................................37

6.1研究结论............................................37

6.2存在问题与挑战......................................38

6.3未来研究方向..