DGIST Scholar: Post-hoc Analysis of Trained Classification Models for Interpretability

Title: Post-hoc Analysis of Trained Classification Models for Interpretability

Citation: Jaemyeong Eo. (2021). Post-hoc Analysis of Trained Classification Models for Interpretability. doi: 10.22677/thesis.200000368542

Subject: Deep Learning, Convolutional Neural Network, Curriculum Learning, Model Interpretability, 심층신경망, 설명 가능한 모델, 커리큘럼 학습

Table Of Contents: Abstract i
List of contents ii
I. INTRODUCTION
1.1 Machine Learning Model Interpretability 1
1.2 Curriculum Learning 2
1.3 Contributions and Overview of Thesis 2
II. BATCH-WISE POST-HOC ANALYSIS FOR IMAGE CLASSIFICATION
2.1 Dataset 3
2.2 Neural Network Model and Experiment Environment 3
2.3 Experiment Design 4
2.4 Result Analysis 4
III. CONCLUSION AND FUTURE WORK
3.1 Concluding remarks 9
3.2 Future work 9
IV. Appendix
4.1 Python code 10
References 13
요약문 14

URI: http://dgist.dcollection.net/common/orgView/200000368542
http://hdl.handle.net/20.500.11750/16732

Show Full Item Record