Model_make

# -*- coding: utf-8 -*-
"""
Created on Mon May 23 15:47:53 2016

@author: kaku
Make CNN structure by using models.Sequential
cnn structure

input:  3 x 18 x 18           
                              core        border_mode                 output
layer1: convolution2d         6 x 5 x 5   valid       sigmoid         3 x 14 x 14
layer2: AveragePooling2d      2 x 2       valid       stride=2        3 x 7 x 7
layer3: convolution2d         12x 4 x 4   valid       sigmoid         3 x 4 x 4
layer4: AveragePooling2d      2 x 2       valid       stride=2        3 x 2 x 2
layer5: Full connected                                                3 x 4 x 1
        output                                        softmax         2 x 1
"""
from keras.models import Sequential
from keras.layers import Dense, Activation, Flatten
from keras.layers import Convolution2D
from keras.layers.convolutional import AveragePooling2D

def creat_model():
    model=Sequential()
    model.add(Convolution2D(6,5,5, border_mode= 'valid',input_shape=(3,18,18)))
    model.add(Activation('sigmoid'))
    model.add(AveragePooling2D(pool_size=(2, 2), strides=(2, 2), border_mode='valid'))
    model.add(Convolution2D(12,4,4, border_mode='valid',input_shape=(3,7,7)))
    model.add(Activation('sigmoid'))
    model.add(AveragePooling2D(pool_size=(2,2), strides=(2,2), border_mode='valid'))
    model.add(Flatten())
    model.add(Dense(2))
    model.add(Activation('sigmoid'))
    
    model.compile(loss='categorical_crossentropy', optimizer='sgd', metrics=['accuracy'])
    return model