Deeplearning-from-scratch

Deeplearning from scratch book: http://www.hanbit.co.kr/store/books/look.php?p_code=B8475831198

---

Deeplearning from scratch(밑바닥부터 시작하는 딥러닝) 책을 공부하면서 정리한 내용입니다.

Ch01: Hello python

1. hungry.py (print)
2. man.py (class)
3. numpy_intro (행렬연산)
4. matplotlib_intro (시각화)

Ch02: Perceptron

1. logic_gate

Ch03: Neural network

1. activation function
2. three layer neuralnet
3. MNIST example and batch

Ch04: Neural network training

1. loss function
2. gradient descent
3. neuralnet training
4. two-layer net

Ch05: Backpropagation

1. layer naive
2. activation layers propagation
3. backpropagation
4. gradient check
5. train neuralnet

Ch06: Techniques

1. optimizer
2. optimizer compare naive
3. optimizer compare MNIST
4. weight initialize, activation histogram
5. weight initialize compare
6. batch norm test
7. overfit weight decay
8. overfit dropout
9. hyperparameter optimization

Ch07: CNN

1. warmingup
2. simple convent
3. train convent
4. gradient check
5. visualize filter
6. apply filter

Ch08: Deep learning

1. deep convent – VGGNet
2. train deepnet – VGGNet
3. misclassified MNIST
4. half float network

Common

위 코드 중에서 자주 사용되는 function, util등을 코드화

dataset

학습에 사용하기 위한 dataset