This is the project of the first homework of Computer Vision course.
Run search_data_generate.py
to do the stochastic search of hyper parameters. You can also modify the search space and search parameters. Data would be generated in the 'save_data' folder.
Run search_data_analysis.py
, and it will show you about the good models.
The best model is also put here in the repo. Link
If you would like to use the model, first clone the repo, then load the model using this code:
from neuralNetworkModel import NeuralNetworkModel
model = NeuralNetworkModel.from_save_data(filepath_to_save_data)
Then, you can happily predict any data you want.
# input should suit the shape of (n, 784)
# if you have images, you would want to reshape them to 1-d vectors
model.predict(x_input)