Achieve some Logistic-Regression solver with numpy. Compare different optimization algorithm in those model, waiting for more optimization algorithm.
Use sklearn.datesets to generate training data.
Logistic Regression with difference optimization algorithm.
RUN EXAMPLE:
- Gauss-Newton iteration (GN):
python logistic.py GN - Gradient Descent (GD):
python logistic.py GD --learning_rate=0.001 --iteration=500 - Stochastic Gradient Descent (SGD):
python logistic.py SGD --learning_rate=0.01 - Mini Batch Gradient Descent (MBGD):
python logistic.py MBGD --learning_rate=0.001 --iteration=50 --batch_size=20