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 | 31 |
Tags
- Classificaion
- Attention
- abstraction
- Set Transformer
- sigmoid
- sentence embedding
- overfitting
- neural net
- NLP
- word2vec
- GNN
- Transformer
- deep learning
- elif
- python practice
- XOR
- Python
- pytorch
- Self-loop attention
- machine learning
Archives
- Today
- Total
목록ReLU (1)
Research Notes

▶ ReLU의 출현 - 신경망의 층이 깊을때, Backpropagation을 사용하였을 경우 학습이 잘 안됨 : Vanishing Gradient(경사 감소 소멸) - sigmoid의 잘못된 사용 ▶ReLU : Rectified Linear Unit - ReLU 함수 : x가 0보다 크면 y=x, x가 0보다 작으면 0 ▶ 다양한 Activation Functions - sigmoid, tanh, ReLU, Leaky ReLU, Maxout, ELU 등 여러 가지의 활성화함수가 존재 ▶ Weight 초기값 정하기 - Vanishing Gradient를 해결할 수 있는 방법 1) ReLU 함수의 사용 2) Weight의 초기값을 잘 정하기 - 초기값을 줄 때 주의점 · 값에 0을 주면 안됨 · RBM(..
Study/Deep Learning
2023. 7. 3. 11:31