grid search improved

.ipynb_checkpoints/Untitled-checkpoint.ipynb

@@ -0,0 +1,6 @@
+{
+ "cells": [],
+ "metadata": {},
+ "nbformat": 4,
+ "nbformat_minor": 2
+}

scnets.py

@@ -9,6 +9,8 @@ from keras.wrappers.scikit_learn import KerasRegressor
 from keras.optimizers import Adam
 import numpy as np
 import matplotlib.pyplot as plt
+from keras.layers import PReLU
+
 
 #function to test performance on testset
 def calc_mre(y_true, y_pred):
@@ -26,35 +28,96 @@ def relerr_loss(y_true, y_pred):
     y_err_f = K.flatten(y_err)
     return K.sum(y_err_f)
 
-def conv1dmodel(in_size=8, out_size=256, ker_size=3):
+
+
+
+
+
+def conv1dmodel(in_size=8, 
+        out_size=256,
+        c1_nf=64,
+        clayers=2,
+        ker_size=3):
     # create model
     model = Sequential()
 
     model.add(Dense(out_size, input_dim=in_size, 
         kernel_initializer='normal',
-        name='first' ))
-    model.add(Activation('relu'))
-    
+        name='first', activation='relu' ))
+   
     model.add(Reshape((4, 64), name='Reshape1'))
     model.add(UpSampling1D(size=2, name='Up1'))
 
-    model.add(Conv1D(filters=64, 
+    model.add(Conv1D(filters=c1_nf, 
         kernel_size=ker_size, strides=1, padding='same', 
         dilation_rate=1, name='Conv1', 
-        kernel_initializer='normal'))
-    model.add(Activation('relu'))
+        kernel_initializer='normal', activation='relu'))
+
+
+    for cl in np.arange(clayers):
+        model.add(Conv1D(filters=32, 
+            kernel_size=ker_size, 
+            strides=1, 
+            padding='same', 
+            dilation_rate=1, 
+            name='Conv'+ str(cl+2),
+            kernel_initializer='normal',
+            activation='relu'))
+
+    model.add(Flatten()) 
+    
+    model.compile(loss=relerr_loss, optimizer='adam', metrics=[calc_mre_K])
+    return model
+
+
+def convprel(in_size=8, 
+        out_size=256,
+        c1_nf=64,
+        clayers=2,
+        ker_size=3):
+    
+    # create model
+    model = Sequential()
 
-    model.add(Conv1D(filters=32, 
+    model.add(Dense(out_size, input_dim=in_size, 
+        kernel_initializer='normal',
+        name='first'))
+    model.add(PReLU(alpha_initializer='zeros', alpha_regularizer=None))
+    model.add(Reshape((4, 64), name='Reshape1'))
+    model.add(UpSampling1D(size=2, name='Up1'))
+
+    model.add(Conv1D(filters=c1_nf, 
         kernel_size=ker_size, strides=1, padding='same', 
-        dilation_rate=1, name='Conv2',
+        dilation_rate=1, name='Conv1', 
         kernel_initializer='normal'))
-    model.add(Activation('relu'))
+    model.add(PReLU(alpha_initializer='zeros', alpha_regularizer=None))
+
 
+    for cl in np.arange(clayers):
+        model.add(Conv1D(filters=32, 
+            kernel_size=ker_size, 
+            strides=1, 
+            padding='same', 
+            dilation_rate=1, 
+            name='Conv'+ str(cl+2),
+            kernel_initializer='normal'))
+        model.add(PReLU(alpha_initializer='zeros', alpha_regularizer=None))
+        
+        
     model.add(Flatten()) 
     
-    model.compile(loss=relerr_loss, optimizer='adam', metrics=['accuracy', calc_mre_K])
+    model.compile(loss=relerr_loss, optimizer='adam', metrics=[calc_mre_K])
     return model
 
+
+
+
+
+
+
+
+
+
 def fullycon( in_size=8, out_size=250, N_hidden=3, N_neurons=250, N_gpus=1):
     """
     Returns a fully-connected model which will take a normalized size vector and return a
@@ -75,10 +138,10 @@ def fullycon( in_size=8, out_size=250, N_hidden=3, N_neurons=250, N_gpus=1):
 
     # Compile model
     if N_gpus == 1:
-        model.compile(loss=relerr_loss, optimizer='adam', metrics=['accuracy'])
+        model.compile(loss=relerr_loss, optimizer='adam', metrics=[calc_mre_K])
     else:
         gpu_list = ["gpu(%d)" % i for i in range(N_gpus)]
-        model.compile(loss=relerr_loss, optimizer='adam', metrics=['accuracy'], context = gpu_list)
+        model.compile(loss=relerr_loss, optimizer='adam', metrics=[calc_mre_K], context = gpu_list)
     return model
 
 #staging area for new models