Merge dev (shenweichen#12)

# Major Changes and Improvements
- Refactor `deepctr.models.MLR`
- Update `deepctr.models.PredictionLayer` to support Layer type activations.
- Add default args to `deepctr.layers.MLP` and `deepctr.layers.LocalActivationUnit`
- Fix bug in `deepctr.layers.DCN` when only use CrossNet.
- Fix bug in get_config of `deepctr.layers.LocalActivationUnit`
- Add many test cases in `tests` folder
- Add continuous integration, test coverage detection and  code quality detection
- Other small changes

Author: Weichen Shen

.gitignore

@@ -1,4 +1,5 @@
 *.h5
+.pytest_cache/
 tests/unused/*
 # Byte-compiled / optimized / DLL files
 __pycache__/

README.md

@@ -1,17 +1,18 @@
 # DeepCTR
 ![dep1](https://img.shields.io/badge/Tensorflow-1.4/1.5/1.6-blue.svg
 )
-[![Python Versions](https://img.shields.io/badge/python-3.6-blue.svg)](https://pypi.org/project/deepctr)
+[![Python Versions](https://img.shields.io/pypi/pyversions/deepctr.svg)](https://pypi.org/project/deepctr)
 [![PyPI Version](https://img.shields.io/pypi/v/deepctr.svg)](https://pypi.org/project/deepctr)
 [![GitHub Issues](https://img.shields.io/github/issues/shenweichen/deepctr.svg
 )](https://github.com/shenweichen/deepctr/issues)
 [![License](https://img.shields.io/badge/license-MIT-brightgreen.svg)](https://github.com/shenweichen/deepctr/blob/master/LICENSE)
 
 [![Documentation Status](https://readthedocs.org/projects/deepctr-doc/badge/?version=latest)](https://deepctr-doc.readthedocs.io/)
 [![Build Status](https://travis-ci.com/shenweichen/DeepCTR.svg?branch=master)](https://travis-ci.com/shenweichen/DeepCTR)
-[![Coverage Status](https://coveralls.io/repos/github/shenweichen/DeepCTR/badge.svg)](https://coveralls.io/github/shenweichen/DeepCTR)
+[![Coverage Status](https://coveralls.io/repos/github/shenweichen/DeepCTR/badge.svg?branch=master)](https://coveralls.io/github/shenweichen/DeepCTR?branch=master)
 [![Codacy Badge](https://api.codacy.com/project/badge/Grade/d4099734dc0e4bab91d332ead8c0bdd0)](https://www.codacy.com/app/wcshen1994/DeepCTR?utm_source=github.com&utm_medium=referral&utm_content=shenweichen/DeepCTR&utm_campaign=Badge_Grade)
 
+
 DeepCTR is a **Easy-to-use**,**Modular** and **Extendible** package of deep-learning based CTR models along with lots of core components layer  which can be used to build your own custom model easily.You can use any complex model with `model.fit()`and`model.predict()` just like any other keras model.And the layers are compatible with tensorflow.Through  `pip install deepctr`  get the package and [**Get Started!**](https://deepctr-doc.readthedocs.io/en/latest/Quick-Start.html)
 
 

deepctr/activations.py

@@ -1,4 +1,4 @@
-from tensorflow.python.keras.layers import Layer, BatchNormalization
+from tensorflow.python.keras.layers import Layer
 from tensorflow.python.keras.initializers import Zeros
 import tensorflow as tf
 
@@ -26,19 +26,20 @@ def __init__(self, axis=-1, epsilon=1e-9, **kwargs):
         self.epsilon = epsilon
         super(Dice, self).__init__(**kwargs)
 
-
     def build(self, input_shape):
         self.alphas = self.add_weight(shape=(input_shape[-1],), initializer=Zeros(
         ), dtype=tf.float32, name=self.name+'dice_alpha')  # name='alpha_'+self.name
         super(Dice, self).build(input_shape)  # Be sure to call this somewhere!
 
     def call(self, inputs, **kwargs):
 
-        inputs_normed = tf.layers.batch_normalization(inputs,axis=self.axis, epsilon=self.epsilon, center=False, scale=False)
+        inputs_normed = tf.layers.batch_normalization(
+            inputs, axis=self.axis, epsilon=self.epsilon, center=False, scale=False)
         x_p = tf.sigmoid(inputs_normed)
         return self.alphas * (1.0 - x_p) * inputs + x_p * inputs
+
     def get_config(self,):
 
-        config = {'axis': self.axis,'epsilon':self.epsilon}
+        config = {'axis': self.axis, 'epsilon': self.epsilon}
         base_config = super(Dice, self).get_config()
-        return dict(list(base_config.items()) + list(config.items()))
+        return dict(list(base_config.items()) + list(config.items()))

deepctr/layers.py

@@ -1,6 +1,6 @@
 from tensorflow.python.keras.layers import Layer,Activation,BatchNormalization
 from tensorflow.python.keras.regularizers import  l2
-from tensorflow.python.keras.initializers import  RandomNormal,Zeros,glorot_normal,glorot_uniform
+from tensorflow.python.keras.initializers import Zeros,glorot_normal,glorot_uniform
 from tensorflow.python.keras import backend as K
 
 import tensorflow as tf
@@ -61,7 +61,7 @@ class AFMLayer(Layer):
 
         - **l2_reg_w** : float between 0 and 1. L2 regularizer strength applied to attention network.
 
-        - **keep_prob** : float between 0 and 1. Fraction of the attention net output units to keep. 
+        - **keep_prob** : float between 0 and 1. Fraction of the attention net output units to keep.
 
         - **seed** : A Python integer to use as random seed.
 
@@ -175,9 +175,10 @@ def call(self, inputs,**kwargs):
 
         if isinstance(self.activation,str):
             output = Activation(self.activation)(x)
+        elif issubclass(self.activation,Layer):
+            output = self.activation()(x)
         else:
-            output = self.activation(x)
-
+            raise ValueError("Invalid activation of MLP,found %s.You should use a str or a Activation Layer Class." % (self.activation))
         output = tf.reshape(output,(-1,1))
 
         return output
@@ -254,7 +255,7 @@ def get_config(self,):
 
 class MLP(Layer):
     """The Multi Layer Percetron
-        
+
       Input shape
         - nD tensor with shape: ``(batch_size, ..., input_dim)``. The most common situation would be a 2D input with shape ``(batch_size, input_dim)``.
 
@@ -268,14 +269,14 @@ class MLP(Layer):
 
         - **l2_reg**: float between 0 and 1. L2 regularizer strength applied to the kernel weights matrix.
 
-        - **keep_prob**: float between 0 and 1. Fraction of the units to keep. 
+        - **keep_prob**: float between 0 and 1. Fraction of the units to keep.
 
         - **use_bn**: bool. Whether use BatchNormalization before activation or not.
 
         - **seed**: A Python integer to use as random seed.
     """
 
-    def __init__(self,  hidden_size, activation,l2_reg, keep_prob, use_bn,seed,**kwargs):
+    def __init__(self,  hidden_size, activation='relu',l2_reg=0, keep_prob=1, use_bn=False,seed=1024,**kwargs):
         self.hidden_size = hidden_size
         self.activation =activation
         self.keep_prob = keep_prob
@@ -338,7 +339,7 @@ class BiInteractionPooling(Layer):
     """Bi-Interaction Layer used in Neural FM,compress the pairwise element-wise product of features into one single vector.
 
       Input shape
-        - A list of 3D tensor with shape:``(batch_size,field_size,embedding_size)``.
+        - A 3D tensor with shape:``(batch_size,field_size,embedding_size)``.
 
       Output shape
         - 3D tensor with shape: ``(batch_size,1,embedding_size)``.
@@ -381,7 +382,7 @@ class OutterProductLayer(Layer):
 
       Output shape
             - 2D tensor with shape:``(batch_size,N*(N-1)/2 )``.
-    
+
       Arguments
             - **kernel_type**: str. The kernel weight matrix type to use,can be mat,vec or num
 
@@ -557,7 +558,7 @@ def call(self, inputs,**kwargs):
         row = []
         col = []
         num_inputs = len(embed_list)
-        num_pairs = int(num_inputs * (num_inputs - 1) / 2)
+        #num_pairs = int(num_inputs * (num_inputs - 1) / 2)
 
 
         for i in range(num_inputs - 1):
@@ -604,7 +605,7 @@ class LocalActivationUnit(Layer):
 
         - **l2_reg**: float between 0 and 1. L2 regularizer strength applied to the kernel weights matrix of attention net.
 
-        - **keep_prob**: float between 0 and 1. Fraction of the units to keep of attention net. 
+        - **keep_prob**: float between 0 and 1. Fraction of the units to keep of attention net.
 
         - **use_bn**: bool. Whether use BatchNormalization before activation or not in attention net.
 
@@ -614,7 +615,7 @@ class LocalActivationUnit(Layer):
         - [Deep Interest Network for Click-Through Rate Prediction](https://arxiv.org/pdf/1706.06978.pdf)
     """
 
-    def __init__(self,hidden_size, activation,l2_reg, keep_prob, use_bn,seed,**kwargs):
+    def __init__(self,hidden_size=(64,32), activation='sigmoid',l2_reg=0, keep_prob=1, use_bn=False,seed=1024,**kwargs):
         self.hidden_size = hidden_size
         self.activation = activation
         self.l2_reg = l2_reg
@@ -663,7 +664,7 @@ def compute_output_shape(self, input_shape):
         return input_shape[1][:2] + (1,)
 
     def get_config(self,):
-        config = {'activation': self.activation,'hidden_size':self.hidden_size, 'l2_reg':self.l2_reg, 'keep_prob':self.keep_prob,'seed': self.seed}
+        config = {'activation': self.activation,'hidden_size':self.hidden_size, 'l2_reg':self.l2_reg, 'keep_prob':self.keep_prob,'use_bn':self.use_bn,'seed': self.seed}
         base_config = super(LocalActivationUnit, self).get_config()
         return dict(list(base_config.items()) + list(config.items()))
 

deepctr/models/__init__.py

@@ -2,7 +2,6 @@
 from .dcn import DCN
 from .mlr import MLR
 from .deepfm import DeepFM
-from .mlr import MLR
 from .nfm import NFM
 from .din import DIN
 from .fnn import FNN

deepctr/models/afm.py

@@ -61,8 +61,6 @@ def AFM(feature_dim_dict, embedding_size=8, use_attention=True, attention_factor
         linear_term = add(linear_term)
     elif len(linear_term) > 0:
         linear_term = linear_term[0]
-    else:
-        linear_term = 0
 
     if len(dense_input) > 0:
         continuous_embedding_list = list(

deepctr/models/dcn.py

@@ -16,7 +16,7 @@
 
 
 def DCN(feature_dim_dict, embedding_size='auto',
-        cross_num=2, hidden_size=[128, 128, ], l2_reg_embedding=1e-5, l2_reg_cross=1e-5, l2_reg_deep=0,
+        cross_num=2, hidden_size=(128, 128, ), l2_reg_embedding=1e-5, l2_reg_cross=1e-5, l2_reg_deep=0,
         init_std=0.0001, seed=1024, keep_prob=1, use_bn=False, activation='relu', final_activation='sigmoid',
         ):
     """Instantiates the Deep&Cross Network architecture.
@@ -49,7 +49,8 @@ def DCN(feature_dim_dict, embedding_size='auto',
     embed_list = [sparse_embedding[i](sparse_input[i])
                   for i in range(len(sparse_input))]
 
-    deep_input = Flatten()(Concatenate()(embed_list))
+    deep_input = Flatten()(Concatenate()(embed_list)
+                           if len(embed_list) > 1 else embed_list[0])
     if len(dense_input) > 0:
         if len(dense_input) == 1:
             continuous_list = dense_input[0]

deepctr/models/deepfm.py

@@ -19,7 +19,7 @@
 
 
 def DeepFM(feature_dim_dict, embedding_size=8,
-           use_fm=True, hidden_size=[128, 128], l2_reg_linear=0.00001, l2_reg_embedding=0.00001, l2_reg_deep=0,
+           use_fm=True, hidden_size=(128, 128), l2_reg_linear=0.00001, l2_reg_embedding=0.00001, l2_reg_deep=0,
            init_std=0.0001, seed=1024, keep_prob=1, activation='relu', final_activation='sigmoid', use_bn=False):
     """Instantiates the DeepFM Network architecture.
 
@@ -61,8 +61,6 @@ def DeepFM(feature_dim_dict, embedding_size=8,
         linear_term = add(linear_term)
     elif len(linear_term) > 0:
         linear_term = linear_term[0]
-    else:
-        linear_term = 0
 
     if len(dense_input) > 0:
         continuous_embedding_list = list(

deepctr/models/din.py

@@ -9,7 +9,7 @@
 
 from tensorflow.python.keras.layers import Input, Dense, Embedding, Concatenate, Reshape
 from tensorflow.python.keras.models import Model
-from tensorflow.python.keras.initializers import RandomNormal, TruncatedNormal
+from tensorflow.python.keras.initializers import RandomNormal
 from tensorflow.python.keras.regularizers import l2
 
 from ..layers import MLP
@@ -30,7 +30,7 @@ def get_input(feature_dim_dict, seq_feature_list, seq_max_len):
 
 
 def DIN(feature_dim_dict, seq_feature_list, embedding_size=8, hist_len_max=16,
-        use_din=True, use_bn=False, hidden_size=[200, 80], activation='relu', att_hidden_size=[80, 40], att_activation=Dice, att_weight_normalization=False,
+        use_din=True, use_bn=False, hidden_size=(200, 80), activation='relu', att_hidden_size=(80, 40), att_activation=Dice, att_weight_normalization=False,
         l2_reg_deep=0, l2_reg_embedding=1e-5, final_activation='sigmoid', keep_prob=1, init_std=0.0001, seed=1024, ):
     """Instantiates the Deep Interest Network architecture.
 

deepctr/models/fnn.py

@@ -17,7 +17,7 @@
 
 
 def FNN(feature_dim_dict, embedding_size=8,
-        hidden_size=[128, 128],
+        hidden_size=(128, 128),
         l2_reg_embedding=1e-5, l2_reg_linear=1e-5, l2_reg_deep=0,
         init_std=0.0001, seed=1024, keep_prob=1,
         activation='relu', final_activation='sigmoid', ):
@@ -58,8 +58,7 @@ def FNN(feature_dim_dict, embedding_size=8,
         linear_term = add(linear_term)
     elif len(linear_term) > 0:
         linear_term = linear_term[0]
-    else:
-        linear_term = 0
+
     #linear_term = add([linear_embedding[i](sparse_input[i]) for i in range(len(feature_dim_dict["sparse"]))])
     if len(dense_input) > 0:
         continuous_embedding_list = list(