2020 기초인공지능 플젝1
- 최초 등록일
- 2022.11.07
- 최종 저작일
- 2020.10
- 29페이지/
어도비 PDF
- 가격 3,000원
![할인쿠폰받기](/images/v4/document/ico_det_coupon.gif)
소개글
"2020 기초인공지능 플젝1"에 대한 내용입니다.
목차
1. Introduction
2. Discussion
3. Conclusion
4. Appendix A : AlexNet CIFAR-10 Classification Implementaiton
5. Appendix B : Transfer Learning Baseline Model (changes from previous)
6. Appendix C : Transfer Learning Model 1 (changes from previous)
7. Appendix D : Transfer Learning Model 2 (changes from previous)
8. Appendix E : Transfer Learning Model 3 (changes from previous)
9. Appendix F : Transfer Learning Model 3 (Improved) (changes)
본문내용
Introduction
Problems and Solution
Question 1: AlexNet with CIFAR-10
(a) Training Setup
Figure1. Final setup for training and prediction of image labels.
Additional features apart from the original AlexNet model would be (1), (2), (3), (4) and (5) as shown
in figure1 above. Apart from (2) and (4) which were implemented from the beginning,
implementations of (1), (3) and (5) gave better results for test accuracy than the accuracy from
previous implementations without these implementations. Details are provided in discussion section.
(1) Data Augmentation
Figure2. Example of data augmentation
All modifications to original images are done in a random manner. In effect, the training model can
be fed images or random transformations, being able to train on multiple modifications of the same
image over multiple epochs. Such implementation is primarily done to increase diversity in data
that the model uses to train on. Furthermore, the model could become more robust to new and
even noisy images with little modifications.
참고 자료
없음