Notice
Recent Posts
Recent Comments
Link
일 | 월 | 화 | 수 | 목 | 금 | 토 |
---|---|---|---|---|---|---|
1 | 2 | 3 | 4 | 5 | ||
6 | 7 | 8 | 9 | 10 | 11 | 12 |
13 | 14 | 15 | 16 | 17 | 18 | 19 |
20 | 21 | 22 | 23 | 24 | 25 | 26 |
27 | 28 | 29 | 30 |
Tags
- Mask Processing
- canny edge detection
- dynamic programming
- eecs 498
- AlexNet
- One-Stage Detector
- two-stage detector
- machine learning
- real-time object detection
- LSTM
- MySQL
- object detection
- CNN
- opencv
- dfs
- Python
- Reinforcement Learning
- image processing
- BFS
- 강화학습
- YoLO
- 백준
- DP
- MinHeap
- C++
- r-cnn
- deep learning
- 딥러닝
- 그래프 이론
- 머신러닝
Archives
- Today
- Total
목록underfitting (1)
JINWOOJUNG

Research Paperhttps://arxiv.org/pdf/2303.01500https://jmlr.org/papers/v15/srivastava14a.htmlhttps://arxiv.org/abs/1512.033850. AbstractDropout은 Neural Network에서 Overfitting을 방지하는 정규화로써 사용되고 있다. 본 논문에서는 Dropout을 초기학습에서 사용함으로써 Underfitting을 완화하는데 사용할 수 있음을 입증한다.미니배치 간의 기울기 방향 분산을 감소전체 데이터셋의 기울기와 미니 배치의 기울기를 정렬초기 학습에서의 Dropout을 통해 SGD의 확률적 특성을 보완하고, 개별 배치가 모델 학습에 미치는 영향을 제한할 수 있다. 이를 기반으로 Dropout을 ..
딥러닝/논문
2024. 12. 30. 15:35