基于增量学习的多尺度钢材微观组织图像分类

doi:10.11896/jsjkx.230500180

摘要/Abstract

摘要： 钢铁材料微观组织决定了钢的力学性能,因此对钢材微观组织的识别十分重要。钢铁材料显微组织图片放大倍数差异大,同种微观组织在不同放大倍数下的形貌也有很大差别,对持续扩充的多尺度钢材微观组织数据集进行分类的难度很大。因此,结合VGG16网络和自组织增量神经网络(Self Organizing Incremental Neural Network,SOINN),构建基于增量学习的多尺度钢材微观组织图像分类模型;同时,提出基于中心距离的交叉熵损失(Cross Entropy Loss based on Center Distance,CELCD)和交叉训练策略,并融合交叉训练、CELCD和Anchor loss克服“灾难性遗忘”问题,实现对钢材微观组织图片数据的持续学习和高效分类。实验比较了不同增量学习方法在旧数据上的分类精度和“遗忘程度”,结果表明,在增量学习后所提方法的预测精度较增量学习前仅下降14.02% 的前提下,在旧数据上的分类精度最高可达80.49%,与上限精度仅相差5.49%,优于其他增量学习方法。

关键词: 钢材微观组织, 增量学习, 灾难性遗忘, 多尺度, 自组织

Abstract: The mechanical properties of steels are closely related to their microstructures,so it is important to identify the microstructures of steels.The magnification of steel micrograph varies greatly,and the morphology of the same microstructure at different magnifications is also different,so the classification of the continuously expanded multi-scale steel microstructure dataset is difficult.In this paper,VGG16 and self-organizing incremental neural network(SOINN) are combined to build a classification model for multiscale steel microstructure dataset based on incremental learning.In addition,the cross entropy loss based on center distance(CELCD) and cross train strategy are proposed.Combining with cross train,CELCD and anchor loss are utilized to solve the problem of “cata-strophic forgetting” and realize the incremental learning and efficient classification for steel micrographs.The classification accuracy and “forgotten degree” of the model are compared.Experimental results show that after incremental learning,the classification accuracy of the proposed method is only 14.02% lower than that before incremental learning,which reaches 80.49% on the old data and only 5.49% lower than the upper bound,which is superior to other incremental learning methods.

Key words: Steel microstructures, Incremental learning, Catastrophic forgetting, Multi-scale, Self-organizing

中图分类号:

TP391.41

曾培益. 基于增量学习的多尺度钢材微观组织图像分类[J]. 计算机科学, 2024, 51(6A): 230500180-8. https://doi.org/10.11896/jsjkx.230500180

ZENG Peiyi. Classification of Multiscale Steel Microstructure Images Based on Incremental Learning[J]. Computer Science, 2024, 51(6A): 230500180-8. https://doi.org/10.11896/jsjkx.230500180

参考文献

[1]AZIMI S M,BRITZ D,ENGSTLER M,et al.Advanced steel microstructural classification by deep learning methods[J].Scientific Reports,2018,8(1):2128-2128.
[2]LI W G,YANG W,ZHAO Y T,et al.A new method to predict mechanical properties for micro-alloyed steels via industrial data and mechanism analysis[J].Journal of Iron and Steel Research International,2019,26(3):230-241.
[3]MAISURADZE M V,RYZHKOV M A,LEBEDEV D I.Microstructure of the Heat Treated Advanced Low Carbon Steel[J].Solid State Phenomena,2021,316:252-257.
[4]LI W G,SHEN J C,XIE L et al.Semi Supervised Learning of Metallographic Data Based on Self-organizing Incremental and Graph Convolution Neural Network[J].Journal of Electronics &Information Technology,2021,43:1-8.
[5]LOSING V,HAMMER B,WERSING H.Incremental on-line learning:A review and comparison of state of the art algorithms[J].Neurocomputing,2018,275:1261-1274.
[6]YU C,ZHOU Q,LI J,et al.Foundation Model Drives Weakly Incremental Learning for Semantic Segmentation[C]//Procee-dings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.2023:23685-23694.
[7]ZHU F,ZHANG X Y,WANG C,et al.Prototype augmentation and self-supervision for incremental learning[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.2021:5871-5880.
[8]CHOWDHURY A,KAUTZ E,YENER B,et al.Image drivenmachine learning methods for microstructure recognition[J].Computational Materials Science,2016,123:176-187.
[9]LUBBERS N,LOOKMAN T,BARROS K.Inferring low-dimensional microstructure representations using convolutional neural networks[J].Physical Review E,2017,96(5):052111.
[10]DECOST B L,FRANCIS T,HOLM E A.Exploring the microstructure manifold:Image texture representations applied to ultrahigh carbon steel microstructures[J].Acta Materialia,2017,133:30-40.
[11]LI W G,SHEN J C,FAN L X et al.Automatic identification of microstructure of iron and steel material based on convolutional neural network[J].Journal of Iron and Steel Research,2020,32(1):33-43.
[12]MALLYA A,LAZEBNIK S.Packnet:Adding multiple tasks to a single network by iterative pruning[C]//Proceeding of the IEEE Conference on Computer Vision and Pattern Recognition.Salt Lake City,Utah:2018:7765-7773.
[13]FERNANDO C,BANARSE D,BLUNDELL C,et al.PathNet:Evolution channels gradient descent in super neural networks[J].arXiv:1701.08734,2017.
[14]RUSU A A,RABINOWITZ N C,DESJARDINS G,et al.Progressive neural networks[J].arXiv:1606.04671,2016.
[15]REBUFFI S A,KOLESNIKOV A,SPERL G,et al.ICaRL:Incremental classifier and representation learning[C]//Proceeding of the IEEE Conference on Computer Vision and Pattern Recognition.Honolulu,Hawaii.2017:5533-5542.
[16]HINTON G,VINYALS O,DEAN J.Distilling the knowledge in a neural network[C]//Proceeding of the Conference and Workshop on Neural Information Processing Systems.Palais des Congrès de Montréal,Canada,2014.
[17]CHAUDHRY A,RANZATO M,ROHRBACH M,et al.Efficient lifelong learning with A-GEM[C]//Proceeding of the International Conference on Learning Representations.Vancouver,Canada,2018.
[18]CASTRO F M,MARÍN-JIMÉNEZ M J,GUIL N,et al.End-to-end incremental learning[C]//Proceeding of the European Conference on Computer Vision.Munich,Germany 2018:233-248.
[19]WU Y,CHEN Y,WANG L,et al.Large scale incrementallearning[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.Long Beach,California,2019:374-382.
[20]ALJUNDI R,LIN M,GOUJAUD B,et al.Gradient based sample selection for online continual learning[J].Advances in Neural Information Processing Systems,2019,32:11816-11825.
[21]WANG Z,SHEN Y.Incremental Learning for Multi-Interest Sequential Recommendation[C]//2023 IEEE 39th International Conference on Data Engineering(ICDE).IEEE,2023:1071-1083.
[22]LI Z,HOIEM D.Learning without forgetting[J].IEEE Tran-sactions on Pattern Analysis and Machine Intelligence,2017,40(12):2935-2947.
[23]KIRKPATRICK J,PASCANU R,RABINOWITZ N,et al.Overcoming catastrophic forgetting in neural networks[J].Proc. Natl. Acad. Sci. USA,2016,114(13):3521-3526.
[24]DHAR P,SINGH R V,PENG K C,et al.Learning withoutmemorizing[C]//Proceeding of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.Long Beach,California,2019:5138-5146.
[25]HOU S,PAN X,LOY C C,et al.Learning a unified classifier incrementally via rebalancing[C]//Proceeding of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.Long Beach,California,2019:831-839.
[26]ZHANG J,ZHANG J,GHOSHS,et al.Class-incremental lear-ning via deep model consolidation[C]//Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision.Snowmass Village,USA.2020:1131-1140.
[27]TAO X,HONG X,CHANG X,et al.Few-shot class-incremental learning[C]//Proceeding of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.2020:12183-12192.
[28]KOHONEN T.Self-organized formation of topologically correct feature maps[J].Biological Cybernetics,1982,43(1):59-69.
[29]MARTINETZ T,SCHULTEN K.A “neural-gas” networklearns topologies[J/OL].Artificial Neural Networks,1991,397-402.https://www.ks.uiuc.edu/Publications/Papers/PDF/MART91B/MART91B.pdf.
[30]MARTINETZ T M,BERKOVICH S G,SCHULTEN K J.Neural Gas network for vector quantization and its application to time-series prediction[J].IEEE Transactions on Neural Networks,1993,4(4):558-569.
[31]MARTINETZ T M,SCHULTEN K J.Topology representing network[J].Neural Networks,1994,7(3):507-522.
[32]GRAY R M.Vector quantization[J].IEEE ASSP Magazine,1984,1(2):4-29.
[33]FRITZKE B.A growing neural gas network learns topologies[J].Advances in Neural Information Processing Systems,1995,7:625-632.
[34]FURAO S,HASEGAWAO.An incremental network for on-line unsupervised classification and topology learning[J].Neural Networks,2006,19(1):90-106.
[35]SIMONYAN K,ZISSERMAN A.Very deep convolutional networks for large-scale image recognition[J].arXiv:1409.1556,2014.
[36]WEN Y,ZHANG K,LI Z,et al.A discriminative feature learning approach for deep face recognition[C]//European Confe-rence on Computer Vision.Cham:Springer,2016:499-515.
[37]KINGMA D,BA J.Adam:A method for stochastic optimization[J].arXiv:1412.6980,2014.

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed