spearmint-tuner.py

# function that is called by spearmint. This calls the classifier and returns the value that needs to be minimized
# you may choose to return either the loss or the error on the validation set.

# we always do this :/
import numpy as np

# import whichever classifier you need: LinearSVM/Softmax

# Uncomment below line for SVM
# from f15ece6504.classifiers import LinearSVM
# Uncomment below line for Softmax
# from f15ece6504.classifiers import Softmax

# see load_cifar10_tvt.py in the hw0/ folder. This code helps you get your data in a single go. 
# This is basically a function that has all the steps you did in the ipython notebook to ready your data
# for the classifier

from load_cifar10_tvt import load_cifar10_train_val

def get_valError(learning_rate,reg):

	# load data
	X_train,y_train,X_val,y_val = load_cifar10_train_val()
	# init the classifier you need

	# Uncomment below line for SVM
	# classifier = LinearSVM()

	# Uncomment below line for Softmax
	# classifier = Softmax()

	# train classifier
	loss_hist = classifier.train(X_train, y_train, learning_rate, reg, num_iters=1500, verbose=True)
	# get validation error
	y_val_pred = classifier.predict(X_val)
        val_accuracy = np.mean(y_val == y_val_pred)
	# return error rate
	return (1 - val_accuracy)

# The main function interfaces the above function with spearmint. 
def main(job_id, params):
    print 'Anything printed here will end up in the output directory for job #%d' % job_id
    print params
    return get_valError(params['learning_rate'], params['reg'])