[1]鲁绪正,蔡恒进,林莉.基于Capsule网络的甲骨文构件识别方法[J].智能系统学报,2020,15(2):243-254.[doi:10.11992/tis.201904069]
LU Xuzheng,CAI Hengjin,LIN Li.Recognition of Oracle Radical based on the Capsule network[J].CAAI Transactions on Intelligent Systems,2020,15(2):243-254.[doi:10.11992/tis.201904069]
点击复制
LU Xuzheng,CAI Hengjin,LIN Li.Recognition of Oracle Radical based on the Capsule network[J].CAAI Transactions on Intelligent Systems,2020,15(2):243-254.[doi:10.11992/tis.201904069]
基于Capsule网络的甲骨文构件识别方法
《智能系统学报》[ISSN 1673-4785/CN 23-1538/TP] 卷:
15
期数:
2020年第2期
页码:
243-254
栏目:
学术论文—自然语言处理与理解
出版日期:
2020-03-05
- Title:
- Recognition of Oracle Radical based on the Capsule network
- Keywords:
- inscriptions on bones or tortoise shells of the Shang Dynasty; Oracle Radical recognition; convolutional neural networks; Capsule network; dynamic routing algorithm; transfer learning; multi-target recognition; image recognition; ?
- 分类号:
- TP319.4
- 摘要:
- 甲骨文作为中国最早的成形文字系统，具有重要的文化和学术价值。研究甲骨文构件和其构形系统是破译未识别的甲骨文的重要方向，但是甲骨文构件的标记工作只能由资深专家来完成，并且需要耗费大量时间和精力。针对这些问题，提出了一种基于Capsule网络和迁移学习的模型OracleNet，可以自动识别并标记甲骨文字形中包含的构件；同时，构建了包含标记的甲骨文字形和构件数据集，用于模型的训练和评估。实验结果显示，OracleNet模型对甲骨文构件的预测精确度达到了60%以上，其中Top5精确度达到了71.56%，验证了模型的有效性。
- Abstract:
- As the earliest shaped character system in China, the inscriptions on bones or tortoise shells of the Shang Dynasty (c. 16th–11th century BC) have important cultural and academic values. The research on the constructional element, i.e., Oracle Radical, and configuration system of the inscriptions on bones or tortoise shells of the Shang Dynasty is a vital direction to identify the unrecognized Oracle Graphics. However, the marking of Radicals can only be done by experienced experts; moreover, it will take a considerable amount of time and effort. To solve these problems, we proposed a model, i.e., OracleNet, based on the Capsule network and transfer learning, which can automatically identify the Oracle Radical contained in a graphic. At the same time, we built a labeled Oracle Graphics dataset and a labeled Radicals dataset, which were used for training and evaluating the model. The experiment showed that the OracleNet had more than 60% precision for recognizing Radicals in a graphic and the Top 5 precision reached 71.56%, which verified validity of the model.
- 参考文献/References:
-
[1] 朱彦民. 从甲骨文说到中国文化自信[J]. 殷都学刊, 2018, 39(3): 23-34
ZHU Yanmin. From the oracle to Chinese culture confidence[J]. Yindu journal, 2018, 39(3): 23-34
[2] 竺海燕. 甲骨構件與甲骨文構形系統研究[D]. 上海: 华东师范大学, 2005.
ZHU Haiyan. Research of the structural element and structural system of the oracle-bone inscriptions[D]. Shanghai: East China Normal University, 2005.
[3] 周新伦, 李锋, 华星城, 等. 甲骨文计算机识别方法研究[J]. 复旦学报(自然科学版), 1996, 35(5): 481-486
ZHOU Xinlun, LI Feng, HUA Xingcheng, et al. A method of Jia Gu Wen recognition based on a two-level classification[J]. Journal of Fudan University (Natural Science), 1996, 35(5): 481-486
[4] 李锋, 周新伦. 甲骨文自动识别的图论方法[J]. 电子科学学刊, 1996, 18(S1): 41-47
LI Feng, ZHOU Xinlun. Recohnition of Jia Gu Wen based on graph theory[J]. Journal of electronics, 1996, 18(S1): 41-47
[5] 李东琦, 刘永革. 基于构件的甲骨文字编码器设计与实现[J]. 科技创新导报, 2010(15): 18
LI Dongqi, LIU Yongge. Design and implementation of radical-based oracle encoder[J]. Science and technology consulting herald, 2010(15): 18
[6] 高峰, 吴琴霞, 刘永革, 等. 基于语义构件的甲骨文模糊字形的识别方法[J]. 科学技术与工程, 2014, 14(30): 67-70, 86
GAO Feng, WU Qinxia, LIU Yongge, et al. Recognition of fuzzy inscription character based on component for bones or tortoise shells[J]. Science technology and engineering, 2014, 14(30): 67-70, 86
[7] 吴琴霞, 栗青生, 高峰. 基于语义构件的甲骨文字库自动生成技术研究[J]. 北京大学学报(自然科学版), 2014, 50(1): 161-166
WU Qinxia, LI Qingsheng, GAO Feng. Study on the technique of automatic generation of oracle characters based on semantic component[J]. Acta Scientiarum Naturalium Universitatis Pekinensis, 2014, 50(1): 161-166
[8] 顾绍通. 基于分形几何的甲骨文字形识别方法[J]. 中文信息学报, 2018, 32(10): 138-142
GU Shaotong. Identification of oracle-bone script fonts based on fractal geometry[J]. Journal of Chinese information processing, 2018, 32(10): 138-142
[9] 李宗焜. 甲骨文字编[M]. 北京: 中华书局, 2012.
[10] SABOUR S, FROSST N, HINTON G E. Dynamic routing between capsules[C]//Proceedings of the 31st Conference on Neural Information Processing Systems. Long Beach, USA, 2017: 3856-3866.
[11] LECUN Y, BOTTOU L, BENGIO Y, et al. Gradient-based learning applied to document recognition[J]. Proceedings of the IEEE, 1998, 86(11): 2278-2324.
[12] KRIZHEVSKY A, SUTSKEVER I, HINTON G E. ImageNet classification with deep convolutional neural networks[C]//Proceedings of the 25th International Conference on Neural Information Processing Systems. Lake Tahoe, USA, 2012: 1097-1105.
[13] HINTON G, SABOUR S, FROSST N. Matrix capsules with EM routing[C]//Proceedings of the 6th International Conference on Learning Representations, ICLR. 2018.
[14] GIRSHICK R, DONAHUE J, DARRELL T, et al. Rich feature hierarchies for accurate object detection and semantic segmentation[C]//Proceedings of 2014 IEEE Conference on Computer Vision and Pattern Recognition. Columbus, USA, 2014: 580-587.
[15] GIRSHICK R. Fast R-CNN[C]//Proceedings of 2015 IEEE International Conference on Computer Vision. Santiago, USA, 2015: 1440-1448.
[16] SRIVASTAVA N, HINTON G, KRIZHEVSKY A, et al. Dropout: a simple way to prevent neural networks from overfitting[J]. The journal of machine learning research, 2014, 15(1): 1929-1958.
[17] GLOROT X, BORDES A, BENGIO Y. Deep sparse rectifier neural networks[C]//Proceedings of the 14th International Conference on Artificial Intelligence and Statistics. Fort Lauderdale, USA, 2011: 315-323.
[18] IOFFE S, SZEGEDY C. Batch normalization: accelerating deep network training by reducing internal covariate shift[EB/OL]. [2018-10-15] https://arxiv.org/abs/1502.03167v1.
[19] KINGMA D P, BA J. Adam: a method for stochastic optimization[EB/OL]. [2018-10-05] https://arxiv.org/abs/1412.6980.
[20] SZEGEDY C, VANHOUCKE V, IOFFE S, et al. Rethinking the inception architecture for computer vision[C]//Proceedings of 2016 IEEE Conference on Computer Vision and Pattern Recognition. Las Vegas, USA, 2016: 2818-2826.
[21] HE Kaiming, ZHANG Xiangyu, REN Shaoqing, et al. Deep residual learning for image recognition[C]//Proceedings of 2016 IEEE Conference on Computer Vision and Pattern Recognition. Las Vegas, USA, 2016: 770-778.
[22] CHOLLET F. Xception: deep learning with depthwise separable convolutions[C]//Proceedings of 2017 IEEE Conference on Computer Vision and Pattern Recognition. Honolulu, USA, 2017: 1251-1258.
[23] PEREZ L, WANG J. The effectiveness of data augmentation in image classification using deep learning[EB/OL]. [2018-09-05] https://arxiv.org/abs/1712.04621.
备注/Memo
收稿日期:2019-05-05。
作者简介:鲁绪正，硕士研究生，主要研究方向为机器学习、深度学习、人工智能;蔡恒进，教授，博士生导师，主要研究方向为区块链技术、人工智能、软件工程、金融信息工程。卓尔智联研究院执行院长，中国科学院深圳先进技术研究院多媒体集成技术研究中心客座研究员，中国人工智能和大数据百人会专家委员，中国通信工业协会区块链专业委员会副主任委员。著作《机器崛起前传——自我意识与人类智慧的开端》获得2017年吴文俊人工智能科学技术奖。发表学术论文100余篇，授权国家专利4项;林莉，硕士研究生，主要研究方向为人工智能
通讯作者:蔡恒进.E-mail:hjcai@whu.edu.cn
更新日期/Last Update:
1900-01-01