"深度学习在时间序列分类中的应用"

版权申诉

5星 · 超过95%的资源 59 浏览量更新于2024-03-05 7 收藏 5.18MB PDF 举报

"时间序列分类：基于深度学习的时间序列分类"是一项致力于利用深度学习技术对时间序列数据进行分类的研究。该研究旨在利用深度学习算法的优势，通过对时间序列数据的特征提取和模式识别，实现对不同分类的时间序列数据进行准确分类。本研究是由POYRAZ UMUT HATIPO ˘GLU在2016年提交给中东技术大学自然与应用科学研究生院的一项硕士学位论文。论文的主要目的是探究如何利用深度学习技朝向时间序列数据进行分类，并且获得了工业工程硕士学位。深度学习是一种模仿人脑神经元网络结构的机器学习算法，它具有强大的特征提取和模式识别能力，特别适用于处理具有复杂结构和大量数据的时间序列。在本研究中，POYRAZ UMUT HATIPO ˘GLU运用深度学习算法，对时间序列数据进行特征提取和分类，以实现对不同类型时间序列数据的准确分类。对于时间序列数据的分类，深度学习算法能够自动学习数据中的特征，并能够通过层层的神经网络进行多层次的抽象和建模，因此能够对时间序列数据进行更加准确和复杂的分类。此外，POYRAZ UMUT HATIPO ˘GLU在论文中还详细探讨了时间序列数据的特点以及深度学习在处理时间序列数据分类中的优势和挑战。通过对深度学习算法在时间序列数据分类中的研究，POYRAZ UMUT HATIPO ˘GLU提出了一种基于深度学习的时间序列分类方法，该方法能够有效处理不同类型的时间序列数据，并取得了较好的分类效果。同时，论文还对比了不同深度学习算法在时间序列数据分类中的性能，并探讨了深度学习算法的参数选择和模型优化等问题。总的来说，POYRAZ UMUT HATIPO ˘GLU的硕士论文"时间序列分类：基于深度学习的时间序列分类"对深度学习技术在时间序列数据分类领域进行了深入研究，提出了一种有效的分类方法，并为相关研究和应用提供了有益的参考和借鉴。该研究为时间序列数据的分类和深度学习技术的应用提供了新的思路和方法，对相关领域的研究和实际应用具有一定的指导意义。

Table 5.14 Classiﬁcation Performances of Proposed and Benchmark Methods . 67

Table 5.15 Total Time Comparison of DTW 1-NN and Proposed Methods . . . 68

Table A.1 Confusion Matrix of ElectricDevices Dataset . . . . . . . . . . . . . 81

Table A.2 Precision and Recall Values of ElectricDevices Dataset . . . . . . . 81

Table A.3 Confusion Matrix of Wafer Dataset . . . . . . . . . . . . . . . . . . 81

Table A.4 Precision and Recall Values of Wafer Dataset . . . . . . . . . . . . 81

Table A.5 Confusion Matrix of Two Patterns Dataset . . . . . . . . . . . . . . 82

Table A.6 Precision and Recall Values of Two Patterns Dataset . . . . . . . . . 82

Table A.7 Confusion Matrix of ProximalPhalanxOutlineCorrect . . . . . . . . 82

Table A.8 Precision and Recall Values of ProximalPhalanxOutlineCorrect . . . 82

Table A.9 Confusion Matrix of ElectricDevices Dataset . . . . . . . . . . . . . 82

Table A.10Precision and Recall Values of ElectricDevices Dataset . . . . . . . 82

Table A.11Confusion Matrix of Wafer Dataset . . . . . . . . . . . . . . . . . . 83

Table A.12Precision and Recall Values of Wafer Dataset . . . . . . . . . . . . 83

Table A.13Confusion Matrix of Two Patterns Dataset . . . . . . . . . . . . . . 83

Table A.14Precision and Recall Values of Two Patterns Dataset . . . . . . . . . 83

Table A.15Confusion Matrix of ProximalPhalanxOutlineCorrect . . . . . . . . 83

Table A.16Precision and Recall Values of ProximalPhalanxOutlineCorrect . . . 83

Table A.17Confusion Matrix of ElectricDevices Dataset . . . . . . . . . . . . . 84

Table A.18Precision and Recall Values of ElectricDevices Dataset . . . . . . . 84

Table A.19Confusion Matrix of Wafer Dataset . . . . . . . . . . . . . . . . . . 84

Table A.20Precision and Recall Values of Wafer Dataset . . . . . . . . . . . . 84

Table A.21Confusion Matrix of Two Patterns Dataset . . . . . . . . . . . . . . 84

Table A.22Precision and Recall Values of Two Patterns Dataset . . . . . . . . . 84

Table A.23Confusion Matrix of ProximalPhalanxOutlineCorrect . . . . . . . . 84

Table A.24Precision and Recall Values of ProximalPhalanxOutlineCorrect . . . 85

xvi

Figure 4.13 Stack autoencoder network diagram . . . . . . . . . . . . . . . . . 43

Figure 4.14 Restricted Boltzmann network diagram . . . . . . . . . . . . . . . 43

Figure 4.15 Joint probability on visibile and hidden units . . . . . . . . . . . . 44

Figure 4.16 Deep Belief Network diagram (reproduced from [56]) . . . . . . . 45

Figure 4.17 Illustrative diagram of proposed methods for the ﬁrst two networks 48

Figure 5.1 A dynamic programming solution (p=720 & Darker regions imply

bigger distances) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54

Figure 5.2 A dynamic programming solution (p=720 & Warping Window Pa-

rameter 15%) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54

Figure 5.3 Some of the correctly-classiﬁed samples of Wafer dataset . . . . . 65

Figure 5.4 Misclassiﬁed samples of Wafer dataset . . . . . . . . . . . . . . . 65

Figure B.1 Training and Validation Set Errors of ElectricalDevices Dataset for

1-Hidden-Layer-SAE architecture . . . . . . . . . . . . . . . . . . . . . . 89

Figure B.2 Training and Validation Set Errors of ElectricalDevices Dataset for

2-Hidden-Layers-SAE architecture . . . . . . . . . . . . . . . . . . . . . 90

Figure B.3 Training and Validation Set Errors of ElectricalDevices Dataset for

3-Hidden-Layers-SAE architecture . . . . . . . . . . . . . . . . . . . . . 90

Figure B.4 Training and Validation Set Errors of ElectricalDevices Dataset for

4-Hidden-Layers-SAE architecture . . . . . . . . . . . . . . . . . . . . . 91

Figure B.5 Training and Validation Set Errors of ElectricalDevices Dataset for

5-Hidden-Layers-SAE architecture . . . . . . . . . . . . . . . . . . . . . 91

Figure B.6 Training and Validation Set Errors of ElectricalDevices Dataset for

6-Hidden-Layers-SAE architecture . . . . . . . . . . . . . . . . . . . . . 92

Figure B.7 Training and Validation Set Errors of ElectricalDevices Dataset for

7-Hidden-Layers-SAE architecture . . . . . . . . . . . . . . . . . . . . . 92

Figure B.8 Training and Validation Set Errors of Wafer Dataset for 1-Hidden-

Layer-SAE architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . 93

Figure B.9 Training and Validation Set Errors of Wafer Dataset for 2-Hidden-

Layers-SAE architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . 93

xix

Figure B.10Training and Validation Set Errors of Wafer Dataset for 3-Hidden-

Layers-SAE architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . 94

Figure B.11Training and Validation Set Errors of Wafer Dataset for 4-Hidden-

Layers-SAE architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . 94

Figure B.12Training and Validation Set Errors of Wafer Dataset for 5-Hidden-

Layers-SAE architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . 95

Figure B.13Training and Validation Set Errors of Wafer Dataset for 6-Hidden-

Layers-SAE architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . 95

Figure B.14Training and Validation Set Errors of Wafer Dataset for 7-Hidden-

Layers-SAE architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . 96

Figure B.15Training and Validation Set Errors of Two Patterns Dataset for 1-

Hidden-Layer-SAE architecture . . . . . . . . . . . . . . . . . . . . . . . 96

Figure B.16Training and Validation Set Errors of Two Patterns Dataset for 2-

Hidden-Layers-SAE architecture . . . . . . . . . . . . . . . . . . . . . . 97

Figure B.17Training and Validation Set Errors of Two Patterns Dataset for 3-

Hidden-Layers-SAE architecture . . . . . . . . . . . . . . . . . . . . . . 97

Figure B.18Training and Validation Set Errors of Two Patterns Dataset for 4-

Hidden-Layers-SAE architecture . . . . . . . . . . . . . . . . . . . . . . 98

Figure B.19Training and Validation Set Errors of Two Patterns Dataset for 5-

Hidden-Layers-SAE architecture . . . . . . . . . . . . . . . . . . . . . . 98

Figure B.20Training and Validation Set Errors of Two Patterns Dataset for 6-

Hidden-Layers-SAE architecture . . . . . . . . . . . . . . . . . . . . . . 99

Figure B.21Training and Validation Set Errors of Two Patterns Dataset for 7-

Hidden-Layers-SAE architecture . . . . . . . . . . . . . . . . . . . . . . 99

Figure B.22Training and Validation Set Errors of ProximalPhalanxOutlineCor-

rect Dataset for 1-Hidden-Layer-SAE architecture . . . . . . . . . . . . . 100

Figure B.23Training and Validation Set Errors of ProximalPhalanxOutlineCor-

rect Dataset for 2-Hidden-Layers-SAE architecture . . . . . . . . . . . . . 100

Figure B.24Training and Validation Set Errors of ProximalPhalanxOutlineCor-

rect Dataset for 3-Hidden-Layers-SAE architecture . . . . . . . . . . . . . 101

Figure B.25Training and Validation Set Errors of ElectricalDevices Dataset for

1-Hidden-Layer-DBN architecture . . . . . . . . . . . . . . . . . . . . . 101

剩余138页未读，继续阅读

机器学习之心

粉丝: 2w+
资源: 1072

"深度学习在时间序列分类中的应用"

深度学习在线分类：流式深度学习中的时间序列分类

时间序列数据集（UCR）

tsai:时间序列时间序列深度学习 Pytorch fastai - 在 Pytorch fastai 中使用时间序列和序列进行最先进的深度学习

深度学习时间序列预测：基于LSTM模型实战指南

计算机毕业设计：基于深度学习的商品销量LSTM时间序列预测(含matplotlib统计图)，保证可靠运行，赠计算机答辩PPT模板

voice_activity_detection:基于深度学习和TensorFlow的语音活动检测

【代码分享】手把手教你：基于深度学习的滚动轴承故障诊断

Python深度时间序列预测实战：深度学习在时间序列建模的应用

Tundil: 基于深度学习的研究生项目聊天机器人

城市交通流量预测：基于深度学习模型的应用研究

最新资源