簡易檢索 / 詳目顯示

回結果列表
研究生： 蔡豐名
Tsai, Feng-Ming
論文名稱： 基於抑制正常激活的監督式異常分類模型
A Supervised Abnormal Classification Model Based on Suppressing Normal Activation
指導教授： 胡敏君
Hu, Min-Chun
口試委員: 王鈺強
Wang, Yu-Chiang
李祈均
Lee, Chi-Chun
學位類別： 碩士
Master
系所名稱： 電機資訊學院 - 資訊工程學系
Computer Science
論文出版年： 2020
畢業學年度： 108
語文別： 英文
論文頁數： 42
中文關鍵詞： 異常分類 、卷積神經網路 、可解釋性人工智慧
外文關鍵詞： Abnormal Classification, Convolutional Neural Network, Explainable AI
相關次數： 點閱：3  下載：0 
分享至:
查詢本校圖書館目錄 查詢臺灣博碩士論文知識加值系統 勘誤回報

    • 近年來，隨著半導體製程工業依照摩爾定律的大量進步，計算機晶片的運算能力也跟著大幅度的成長，用深度卷積神經網路進行影像辨識變成一個可行的解決方案，目前卷積神經網路應用在許多電腦視覺領域都得到了最佳的研究成果，例如影像辨識、物件偵測、語意分割、姿態估計、臉部辨識等等。在這篇論文中，我們將關注在最為廣泛應用的影像分類問題，傳統的做法中，都需要人類專家花費大量時間去做檢查工作，而深度卷積神經網路不需要人工的介入，可以僅靠著標注好的影像資料就提取出分類所需要的影像特徵並且超越過去所有傳統的電腦視覺技術。在眾多的分類任務中，某些任務存在著不具特徵的類別，意味著這個類別在直覺上不應該具有特徵用來判斷這個類別，比如水泥地路面是否有裂縫的二元任務當中，正常無裂縫的路面就屬於不具特徵的類別，但是當我們直接使用遷移式學習進行這個二元分類任務時，卷積神經網路模型會設法去擬合那些正常的路面影像，進而從正常的影像中學習到不合理的正常特徵，換句話說，直接訓練出來的模型會根據學到的正常特徵來判斷一張路面影像為正常，但我們期許模型判斷一張路面影像為正常是根據影像中沒有裂縫的缺陷特徵。我們提出許多方法來解決這個問題且最後提出一種很簡單的可以套用到大部分卷積神經網路架構上的方法，我們的方法可以學習到更合理且更具解釋性的特徵，並且對於缺陷特徵更加的靈敏，此外，我們的方法對於簡單快速的對抗式攻擊有較高的魯棒性，最後我們在多個資料集呈現我們的方法結果。

    Recently, the computing power of chips grows exponentially due to the massive improvement in semiconductor industry according to Moore’s Law. Deep convolutional neural network (CNN) becomes a practical solution to image recognition. Nowadays, CNN is the state-of-the-art architecture of many computer vision application fields, for example, image recognition, object detection, semantic segmentation, body pose estimation, face recognition and so on. In this thesis, we focus on the most commonly used application: image classification. Traditionally, inspection is a time-consuming job for human experts. Deep CNN can extract the feature of images by the labeled data without any manual interventions and outperforms the traditional computer vision technique. Of all the classification tasks, there are some cases that contain non-featured class which means there is no intuitive feature for this class. For example, in the binary classification of detecting crack on concrete road, the normal road surface should not have any feature. However, when we naïvely apply transfer learning on the binary classification task, the CNN model would fit the normal images and learn unreasonable features. That is, naïve CNN model learns normal road image features to predict an image be normal. Instead, we expect the CNN model predict an image be normal according to it has no defect features. We present many methods to handle this issue and a simple new approach that can easily implement in most of all CNN architectures. We show that our approach learns more reasonable and explainable features and is more sensitive to defect images. Further, our experiment shows that our approach is more robust to fast and simple adversarial attack. Also, we apply our approach on many datasets to demonstrate our contribution.

    Abstract i Contents iii List of Figures v List of Tables vii Chapter 1 Introduction 1 1.1 Background 1 1.2 Problem Motivation 2 1.3 Related Work 2 1.4 Organization of the Thesis 3 Chapter 2 Techniques and Materials 4 2.1 Convolution Neural Network 4 2.2 CNN Visualization 5 2.3 Datasets 6 2.4 Problem Description 6 2.5 Definitions and Notations 7 2.6 Handmade Positive Images 8 Chapter 3 Methodology 11 3.1 Method 1: Tune Prediction Threshold 11 3.2 Method 2: Directly Trim NAL 11 3.3 Method 3: Train by Soft Label 13 3.4 Method 4: Shift Sigmoid Function 14 3.5 Method 5: Add NAL Limitation Loss 16 3.6 Method 6: A Simple New Approach 18 3.6.1 Discussion of Offset in New Approach 18 Chapter 4 Experiment 23 4.1 Adversarial Attack Robustness Test 23 4.2 Discussion of Offset in New Approach 25 4.2.1 Mango Quality Classification 25 4.2.2 Apple Leaves Disease Detection 27 4.2.3 Industrial Optical Inspection Detection 30 4.2.4 Chest X-rays Mass Detection 32 4.3 New Approach on Other Datasets 34 4.4 Suppress Abnormal Activation 35 Chapter 5 Conclusion and Future Work 38 References 39

    1. Y. LeCun, B. Boser, J. S. Denker, D. Henderson, R. E. Howard, W. Hubbard, L. D. Jackel, “Backpropagation Applied to Handwritten Zip Code Recognition,” Neural Computation, vol. 1, no. 4, pp. 541-551, 1989.
    2. Y. Lecun, L. Bottou, Y. Bengio and P. Haffner, "Gradient-based learning applied to document recognition," in Proceedings of the IEEE, vol. 86, no. 11, pp. 2278-2324, 1998.
    3. T. Hahn, M. Wieler, “Weakly Supervised Learning for Industrial Optical Inspection,” 29th Annual Symposium of the German Association for Pattern Recognition, 2007.
    4. J. Deng, W. Dong, R. Socher, L.-J. Li, K. Li and L. Fei-Fei, “ImageNet: A Large-Scale Hierarchical Image Database,” IEEE Computer Vision and Pattern Recognition (CVPR), 2009.
    5. A. Krizhevsky, I. Sutskever, and G. E. Hinton, “ImageNet classification with deep convolutional neural networks,” In Proceedings of the 25th International Conference on Neural Information Processing Systems (NIPS), vol. 1, pp. 1097–1105, 2012.
    6. K. Simonyan, A. Vedaldi, A. Zisserman, “Deep inside convolutional networks: Visualising image classification models and saliency maps.” arXiv preprint arXiv:1312.6034, 2013.
    7. K. Song, Y. Yan, “A noise robust method based on completed local binary patterns for hot-rolled steel strip surface defects,” Applied Surface Science, vol. 285, pp. 858-864, 2013.
    8. C. Szegedy, W. Zaremba, I. Sutskever, J. Bruna, D. Erhan, I. Goodfellow, R. Fergus, “Intriguing properties of neural networks,” International Conference on Learning Representations (ICLR), 2014.
    9. I. Goodfellow, J. Shlens, C. Szegedy, “Explaining and Harnessing Adversarial Examples,” arXiv 1412.6572, 2014.
    10. K. Diederik, B. Jimmy, “Adam: A Method for Stochastic Optimization,” International Conference on Learning Representations, 2014.
    11. M. D. Zeiler, R. Fergus, “Visualizing and Understanding Convolutional Networks,” ECCV, Lecture Notes in Computer Science, vol. 8689, 2014.
    12. C. Szegedy, W. Liu, Y. Jia, P. Sermanet, S. Reed, D. Anguelov, D. Erhan, V. Vanhoucke, A. Rabinovich, “Going Deeper with Convolutions,” Computer Vision and Pattern Recognition (CVPR), 2015.
    13. F. Chollet, “Keras”, GitHub repository, https://github.com/fchollet/keras, 2015.
    14. K. Simonyan, A. Zisserman, “Very Deep Convolutional Networks for Large-Scale Image Recognition,” International Conference on Learning Representations (ICLR), 2015.
    15. M Abadi et. al., “TensorFlow: Large-scale machine learning on heterogeneous systems,” Software available from tensorflow.org, 2015.
    16. B. Zhou, A. Khosla, A. Lapedriza, A. Oliva, A. Torralba, “Learning Deep Features for Discriminative Localization,” IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 2921-2929, 2016.
    17. D. Weimer, B. Scholz-Reiter, M. Shpitalni, “Design of deep convolutional neural network architectures for automated feature extraction in industrial inspection,” CIRP Annals, vol. 65, no. 1, pp.417-420, 2016.
    18. J. T. A. Andrews, E. J. Morton, L. D. Griffin, “Detecting anomalous data using auto-encoders,” International Journal of Machine Learning and Computing (IJMLC), vol. 6, no. 1, 2016.
    19. K. He, X. Zhang, S. Ren and J. Sun, "Deep Residual Learning for Image Recognition," IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 770-778, 2016.
    20. S. F.-Roohi, S. Hajizadeh, A. Núñez, R. Babuska, B. De Schutter, “Deep convolutional neural networks for detection of rail surface defects,” International Joint Conference on Neural Networks (IJCNN), pp. 2584-2589, 2016.
    21. C. Szegedy, S. Ioffe, V. Vanhoucke, A. A. Alemi, “Inception-v4, inception-ResNet and the impact of residual connections on learning,” In Proceedings of the Thirty-First AAAI Conference on Artificial Intelligence (AAAI), pp. 4278-4284, 2017.
    22. F. Chollet, “Xception: Deep Learning with Depthwise Separable Convolutions,” IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 1800-1807, 2017.
    23. K.C.L. Wong, A. Karargyris, T. Syeda-Mahmood, M. Moradi, “Building Disease Detection Algorithms with Very Small Numbers of Positive Samples,” Medical Image Computing and Computer Assisted Intervention (MICCAI), vol. 10435, 2017.
    24. M. Zhang, J. Wu, H. Lin, P. Yuan, Y. Song, “The Application of One-Class Classifier Based on CNN in Image Defect Detection,” Procedia Computer Science, vol. 114, pp. 341-348, 2017.
    25. R. R. Selvaraju, M. Cogswell, A. Das, R. Vedantam, D. Parikh, D. Batra, “Grad-CAM: Visual Explanations From Deep Networks via Gradient-Based Localization,” The IEEE International Conference on Computer Vision (ICCV), pp. 618-626, 2017
    26. G. Huang, Z. Liu, L. V. D. Maaten, K. Q. Weinberger, “Densely Connected Convolutional Networks,” IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 2261-2269, 2017.
    27. X. Wang, Y. Peng, L. Lu, Z. Lu, M. Bagheri, R. M. Summers, “ChestX-ray8: Hospital-scale Chest X-ray Database and Benchmarks on Weakly-Supervised Classification and Localization of Common Thorax Diseases,” IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2017
    28. Y. Zhiyang, W. Xiaojun, G. Xiaodong, “Fully Convolutional Networks for Surface Defect Inspection in Industrial Environment,” pp. 417-426, 2017.
    29. A. Chattopadhay, A. Sarkar, P. Howlader, V. N. Balasubramanian, “Grad-CAM++: Generalized Gradient-Based Visual Explanations for Deep Convolutional Networks,” IEEE Winter Conference on Applications of Computer Vision (WACV), pp. 839-847, 2018.
    30. C. Gao, J. Zhou, W. K. Wong, T. Gao, “Woven Fabric Defect Detection Based on Convolutional Neural Network for Binary Classification,” International Conference on Artificial Intelligence on Textile and Apparel (AITA), Artificial Intelligence on Fashion and Textile, vol. 849, pp. 307-313, 2018.
    31. D. Racki, D. Tomazevic, D. Skocaj, “A Compact Convolutional Neural Network for Textured Surface Anomaly Detection,” IEEE Winter Conference on Applications of Computer Vision (WACV), pp. 1331-1339, 2018.
    32. J. Hu, L. Shen, G. Sun, “Squeeze-and-Excitation Networks,” IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 7132-7141, 2018.
    33. J. Li, Z. Su, J. Geng, Y. Yin, “Real-time Detection of Steel Strip Surface Defects Based on Improved YOLO Detection Network,” IFAC-PapersOnLine, vol. 51, no. 21, pp. 76-81, 2018.
    34. M. Haselmann, D. P. Gruber, P. Tabatabai, “Anomaly Detection Using Deep Learning Based Image Completion,” IEEE International Conference on Machine Learning and Applications (ICMLA), pp. 1237-1242, 2018.
    35. S. Akcay, A. A.-Abarghouei, T. P. Breckon, “GANomaly: Semi-supervised Anomaly Detection via Adversarial Training,” Asian Conference on Computer Vision (ACCV), vol. 11363, 2018.
    36. T. Wang, Y. Chen, M. Qiao, “A fast and robust convolutional neural network-based defect detection model in product quality control,” The International Journal of Advanced Manufacturing Technology, vol. 94, pp. 3465–3471, 2018.
    37. W. Silva, D. S. de Lucena, “Concrete Cracks Detection Based on Deep Learning Image Classification,” The Eighteenth International Conference of Experimental Mechanics, 2018.
    38. X. Tao, D. Zhang, W. Ma, X. Liu, D. Xu, “Automatic Metallic Surface Defect Detection and Recognition with Convolutional Neural Networks,” Applied Sciences, vol. 8, pp. 1575, 2018.
    39. D. Meijer, L. Scholten, F. Clemens, A. Knobbe, “A defect classification methodology for sewer image sets with convolutional neural networks,” Automation in Construction, vol. 104, pp. 281-298, 2019.
    40. J. Irvin, P. Rajpurkar, K. Michael, Y. Yifan, S. C.-Ilcus, C. Chris, H. Marklund, B. Haghgoo, R. Ball, K. Shpanskaya, J. Seekins, D. A. Mong, S. S. Halabi, J. K. Sandberg, R. Jones, D. B. Larson, C. P. Langlotz, B. N. Patel, M. P. Lungren, A. Y. Ng, “CheXpert: A Large Chest Radiograph Dataset with Uncertainty Labels and Expert Comparison,” In AAAI Conference on Artificial Intelligence, 2019.
    41. L.-F. Li, W.-F. Ma, L. Li, C. Lu, “Research on Detection Algorithm for Bridge Cracks Based on Deep Learning,” Acta Automatica Sinica, vol. 45, no. 9, pp. 1727-1742, 2019.
    42. M. M. M. Islam, J.-M. Kim, “Vision-Based Autonomous Crack Detection of Concrete Structures Using a Fully Convolutional Encoder–Decoder Network,” Sensors, vol. 19, pp. 4251, 2019.
    43. M. Tan, Q. V. Le, “EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks,” International Conference on Machine Learning (ICML), pp. 6105-6114, 2019.
    44. Paszke et. al., “PyTorch: An Imperative Style, High-Performance Deep Learning Library,” Advances in Neural Information Processing Systems 32, pp. 8024-8035, 2019.
    45. Q. Zou, Z. Zhang, Q. Li, X. Qi, Q. Wang and S. Wang, “DeepCrack: Learning Hierarchical Convolutional Features for Crack Detection,” IEEE Transactions on Image Processing, vol. 28, no. 3, pp. 1498-1512, 2019.
    46. R. Chalapathy, S. Chawla, “Deep Learning for Anomaly Detection: A Survey,” arXiv 1901.03407, 2019.
    47. S. Cheon, H. Lee, C. O. Kim, S. H. Lee, “Convolutional Neural Network for Wafer Surface Defect Classification and the Detection of Unknown Defect Class,” IEEE Transactions on Semiconductor Manufacturing, vol. 32, no. 2, pp. 163-170, 2019.
    48. X. Hongyan, S. Xiu, W. Yi, C. Huaiyu, C. Kerang, C. Xiaodong, “Automatic Bridge Crack Detection Using a Convolutional Neural Network,” Applied Sciences, vol. 9, pp. 2867, 2019.
    49. Y. Gao, “A semi-supervised convolutional neural network-based method for steel surface defect recognition,” Robotics and Computer-Integrated Manufacturing, vol. 61, 2019.
    50. R. Thapa, N. Snavely, S. Belongie, A. Khan, “The Plant Pathology 2020 challenge dataset to classify foliar disease of apples,” arXiv 2004.11958, 2020.

    QR CODE