andyccs / cifar10 Goto Github PK

We can't plot a graph in EC2, It would be nice to have an argument to do that when running these python script from command line. Instead, we should output the result (a list of accuracy and losses) to the terminal so that we can copy and paste to local desktop and plot the result locally.

I think I made some fundamental mistake in my codes. No matter what technique I use to train my model, it always stuck at 10%. It could be because I mess up with the way I extract data to matrices in images_to_matrices.py.

My intuition tell me that, even if we use simpler techniques such as logistic regression and multinomial logistic regression, we should get something better than 10%. So before we try to test something more advance like convolution neural network, we should get those easier method to work first, at least not 10%.

Using vim to change the script in terminal (when using AWS EC2) is such a pain. Can we have command line arguments to change those experiments that we want to play around and experiment with?

In the file images_to_matrices.py, we still hard code some constants

  image_size = 32
  pixel_depth = 255.0
  color_channel = 3

We should use constants from constants.py instead

Alright, we gonna record all classifier that we have trained here. Please use the following format:

1. How you define your model?
2. How you define your loss function?
3. Parameters
4. Figures
5. Extra Notes (optional)