Tathagatd96

def load_dataset():
    url = 'http://deeplearning.net/data/mnist/mnist.pkl.gz'
    filename = 'mnist.pkl.gz'
    if not os.path.exists(filename):
        print("Downloading MNIST dataset...")
        urlretrieve(url, filename)
    with gzip.open(filename, 'rb') as f:
        data = pickle.load(f)
    X_train, y_train = data[0]
    X_val, y_val = data[1]

Tathagatd96

#Conv Net Structure

batch_size=100
output_size=10
data_size=(None,1,28,28)
input_var = T.tensor4(name='inputs')
target_var =T.ivector(name='targets')


net = {}

Tathagatd96

#Input layer:
net['data'] = lasagne.layers.InputLayer(data_size, input_var=input_var)

#Convolution + Pooling + Normalization 
net['conv1'] = lasagne.layers.Conv2DLayer(net['data'], num_filters=6, filter_size=3)
net['pool1'] = lasagne.layers.Pool2DLayer(net['conv1'], pool_size=2)

net['conv2'] = lasagne.layers.Conv2DLayer(net['pool1'], num_filters=10, filter_size=4)
net['pool2'] = lasagne.layers.Pool2DLayer(net['conv2'], pool_size=2)

Tathagatd96

###Defining the cost function and the update rule
#Define hyperparameters. These could also be symbolic variables 
lr = 1e-2
weight_decay = 1e-5


#Loss function: mean cross-entropy
prediction = lasagne.layers.get_output(net['out'])
loss = lasagne.objectives.categorical_crossentropy(prediction, target_var)
loss = loss.mean()

Tathagatd96

#Get the update rule for Stochastic Gradient Descent
params = lasagne.layers.get_all_params(net['out'], trainable=True)
updates=lasagne.updates.adam(loss,params)

Tathagatd96

###Compiling the training and testing functions
train_fn = theano.function([input_var, target_var], loss, updates=updates)

test_prediction = lasagne.layers.get_output(net['out'], deterministic=True)
test_loss = lasagne.objectives.categorical_crossentropy(test_prediction,
                                                        target_var)
test_loss = test_loss.mean()
test_acc = T.mean(T.eq(T.argmax(test_prediction, axis=1), target_var),
                  dtype=theano.config.floatX)

Tathagatd96

###Training the model
#Run the training function per mini-batches.
n_examples =  X_train.shape[0]
n_batches = n_examples / batch_size
epochs=50

for epoch in xrange(epochs):
    for batch in xrange(n_batches):
        x_batch =  X_train[batch*batch_size: (batch+1) * batch_size]
        y_batch = y_train[batch*batch_size: (batch+1) * batch_size]

Tathagatd96

# -*- coding: utf-8 -*-
"""
Created on Mon Dec 04 17:59:48 2017

@author: Tathagat Dasgupta
"""

# -*- coding: utf-8 -*-
"""
Created on Sat Dec 02 23:56:30 2017

Tathagatd96

import numpy as np
import pandas as pd
import tensorflow as tf
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import MinMaxScaler

df=pd.read_csv("cal_housing_clean.csv")
print(df.describe()) #to understand the dataset

Tathagatd96

y_val= df["medianHouseValue"]
x_data=df.drop("medianHouseValue",axis=1)


X_train, X_eval,y_train,y_eval=train_test_split(x_data,y_val,test_size=0.3,random_state=101)