def test_input_missing():
input_node1 = Input()
input_node2 = Input()
output_node1 = MLPInteraction()(input_node1)
output_node2 = MLPInteraction()(input_node2)
output_node = ConcatenateInteraction()([output_node1, output_node2])
output_node = RatingPredictionOptimizer()(output_node)
with pytest.raises(ValueError) as info:
graph_module.HyperGraph(input_node1, output_node)
assert 'A required input is missing for HyperModel' in str(info.value)
def test_input_output_disconnect():
input_node1 = Input()
output_node = input_node1
_ = MLPInteraction()(output_node)
input_node = Input()
output_node = input_node
output_node = MLPInteraction()(output_node)
output_node = RatingPredictionOptimizer()(output_node)
with pytest.raises(ValueError) as info:
graph_module.HyperGraph(input_node1, output_node)
assert 'Inputs and outputs not connected.' in str(info.value)
def test_graph_basics():
input_node = Input(shape=(30, ))
output_node = input_node
output_node = MLPInteraction()(output_node)
output_node = RatingPredictionOptimizer()(output_node)
graph = graph_module.PlainGraph(input_node, output_node)
model = graph.build_keras_graph().build(hp_module.HyperParameters())
assert model.input_shape == (None, 30)
assert model.output_shape == (None, )
def build_mlp(user_num, item_num):
input = Input(shape=[2])
user_emb_mlp = LatentFactorMapper(feat_column_id=0,
id_num=user_num,
embedding_dim=64)(input)
item_emb_mlp = LatentFactorMapper(feat_column_id=1,
id_num=user_num,
embedding_dim=64)(input)
output = MLPInteraction()([user_emb_mlp, item_emb_mlp])
output = RatingPredictionOptimizer()(output)
model = RPRecommender(inputs=input, outputs=output)
return model
def build_gmf(user_num, item_num):
input = Input(shape=[2])
user_emb = LatentFactorMapper(column_id=0,
num_of_entities=user_num,
embedding_dim=64)(input)
item_emb = LatentFactorMapper(column_id=1,
num_of_entities=item_num,
embedding_dim=64)(input)
output = InnerProductInteraction()([user_emb, item_emb])
output = RatingPredictionOptimizer()(output)
model = RPRecommender(inputs=input, outputs=output)
return model
def build_gmf(user_num, item_num):
input = Input(shape=[2])
user_emb = LatentFactorMapper(feat_column_id=0,
id_num=user_num,
embedding_dim=64)(input)
item_emb = LatentFactorMapper(feat_column_id=1,
id_num=item_num,
embedding_dim=64)(input)
output = ElementwiseInteraction(elementwise_type="innerporduct")(
[user_emb, item_emb])
output = RatingPredictionOptimizer()(output)
model = RPRecommender(inputs=input, outputs=output)
return model
def build_autorec(user_num, item_num):
input = Input(shape=[2])
user_emb_1 = LatentFactorMapper(column_id=0,
num_of_entities=user_num,
embedding_dim=64)(input)
item_emb_1 = LatentFactorMapper(column_id=1,
num_of_entities=item_num,
embedding_dim=64)(input)
user_emb_2 = LatentFactorMapper(column_id=0,
num_of_entities=user_num,
embedding_dim=64)(input)
item_emb_2 = LatentFactorMapper(column_id=1,
num_of_entities=item_num,
embedding_dim=64)(input)
output = HyperInteraction()(
[user_emb_1, item_emb_1, user_emb_2, item_emb_2])
output = RatingPredictionOptimizer()(output)
model = RPRecommender(inputs=input, outputs=output)
return model
def build_neumf(user_num, item_num):
input = Input(shape=[2])
user_emb_gmf = LatentFactorMapper(column_id=0,
num_of_entities=user_num,
embedding_dim=64)(input)
item_emb_gmf = LatentFactorMapper(column_id=1,
num_of_entities=item_num,
embedding_dim=64)(input)
innerproduct_output = InnerProductInteraction()(
[user_emb_gmf, item_emb_gmf])
user_emb_mlp = LatentFactorMapper(column_id=0,
num_of_entities=user_num,
embedding_dim=64)(input)
item_emb_mlp = LatentFactorMapper(column_id=1,
num_of_entities=item_num,
embedding_dim=64)(input)
mlp_output = MLPInteraction()([user_emb_mlp, item_emb_mlp])
output = RatingPredictionOptimizer()([innerproduct_output, mlp_output])
model = RPRecommender(inputs=input, outputs=output)
return model
# load dataset
##Netflix Dataset
# dataset_paths = ["./examples/datasets/netflix-prize-data/combined_data_" + str(i) + ".txt" for i in range(1, 5)]
# data = NetflixPrizePreprocessor(dataset_paths)
# Step 1: Preprocess data
movielens = MovielensPreprocessor()
train_X, train_y, val_X, val_y, test_X, test_y = movielens.preprocess()
train_X_categorical = movielens.get_x_categorical(train_X)
val_X_categorical = movielens.get_x_categorical(val_X)
test_X_categorical = movielens.get_x_categorical(test_X)
user_num, item_num = movielens.get_hash_size()
# Step 2: Build the recommender, which provides search space
# Step 2.1: Setup mappers to handle inputs
input = Input(shape=[2])
user_emb_gmf = LatentFactorMapper(column_id=0,
num_of_entities=user_num,
embedding_dim=64)(input)
item_emb_gmf = LatentFactorMapper(column_id=1,
num_of_entities=item_num,
embedding_dim=64)(input)
user_emb_mlp = LatentFactorMapper(column_id=0,
num_of_entities=user_num,
embedding_dim=64)(input)
item_emb_mlp = LatentFactorMapper(column_id=1,
num_of_entities=item_num,
embedding_dim=64)(input)
# Step 2.2: Setup interactors to handle models
innerproduct_output = InnerProductInteraction()([user_emb_gmf, item_emb_gmf])
criteo = CriteoPreprocessor(
) # the default arguments are setup to preprocess the Criteo example dataset
train_X, train_y, val_X, val_y, test_X, test_y = criteo.preprocess()
train_X_numerical, train_X_categorical = criteo.get_x_numerical(
train_X), criteo.get_x_categorical(train_X)
val_X_numerical, val_X_categorical = criteo.get_x_numerical(
val_X), criteo.get_x_categorical(val_X)
test_X_numerical, test_X_categorical = criteo.get_x_numerical(
test_X), criteo.get_x_categorical(test_X)
numerical_count = criteo.get_numerical_count()
categorical_count = criteo.get_categorical_count()
hash_size = criteo.get_hash_size()
# Step 2: Build the recommender, which provides search space
# Step 2.1: Setup mappers to handle inputs
dense_input_node = Input(shape=[numerical_count])
sparse_input_node = Input(shape=[categorical_count])
dense_feat_emb = DenseFeatureMapper(num_of_fields=numerical_count,
embedding_dim=2)(dense_input_node)
sparse_feat_emb = SparseFeatureMapper(num_of_fields=categorical_count,
hash_size=hash_size,
embedding_dim=2)(sparse_input_node)
# Step 2.2: Setup interactors to handle models
attention_output = SelfAttentionInteraction()(
[dense_feat_emb, sparse_feat_emb])
bottom_mlp_output = MLPInteraction()([dense_feat_emb])
top_mlp_output = MLPInteraction()([attention_output, bottom_mlp_output])
# Step 2.3: Setup optimizer to handle the target task
output = CTRPredictionOptimizer()(top_mlp_output)
level=logging.DEBUG,
format='%(asctime)s - %(name)s - %(levelname)s - %(message)s')
logger = logging.getLogger(__name__)
# load dataset
mini_criteo = np.load("./examples/datasets/criteo/criteo_2M.npz")
# TODO: preprocess train val split
train_X = [
mini_criteo['X_int'].astype(np.float32),
mini_criteo['X_cat'].astype(np.float32)
]
train_y = mini_criteo['y']
val_X, val_y = train_X, train_y
# build the pipeline.
dense_input_node = Input(shape=[13])
sparse_input_node = Input(shape=[26])
dense_feat_emb = DenseFeatureMapper(num_of_fields=13,
embedding_dim=2)(dense_input_node)
# TODO: preprocess data to get sparse hash_size
sparse_feat_emb = SparseFeatureMapper(num_of_fields=26,
hash_size=[
1444, 555, 175781, 128509, 306, 19,
11931, 630, 4, 93146, 5161, 174835,
3176, 28, 11255, 165206, 11, 4606,
2017, 4, 172322, 18, 16, 56456, 86,
43356
],
embedding_dim=2)(sparse_input_node)
# logging setting
logging.basicConfig(
level=logging.INFO,
format='%(asctime)s - %(name)s - %(levelname)s - %(message)s')
logging.basicConfig(
level=logging.DEBUG,
format='%(asctime)s - %(name)s - %(levelname)s - %(message)s')
logger = logging.getLogger(__name__)
# load dataset
criteo = CriteoPreprocessor(
) # automatically set up for preprocessing the Criteo dataset
train_X, train_y, val_X, val_y, test_X, test_y = criteo.preprocess()
# build the pipeline.
input = Input(shape=[criteo.get_categorical_count()])
user_emb_gmf = LatentFactorMapper(column_id=0,
num_of_entities=10000,
embedding_dim=64)(input)
item_emb_gmf = LatentFactorMapper(column_id=1,
num_of_entities=10000,
embedding_dim=64)(input)
user_emb_mlp = LatentFactorMapper(column_id=0,
num_of_entities=10000,
embedding_dim=64)(input)
item_emb_mlp = LatentFactorMapper(column_id=1,
num_of_entities=10000,
embedding_dim=64)(input)
innerproduct_output = InnerProductInteraction()([user_emb_gmf, item_emb_gmf])
mlp_output = MLPInteraction()([user_emb_mlp, item_emb_mlp])