def test_merge_method_seq_concat(self):
bx1 = BLayer.Input(shape=(10, ))
bx1_1 = BLayer.Input(shape=(10, ))
bx2 = BLayer.Input(shape=(10, ))
by1 = BLayer.Dense(12, activation="sigmoid")(bx1)
bbranch1_node = BModel(bx1, by1)(bx1_1)
bbranch2 = BSequential()
bbranch2.add(BLayer.Dense(12, input_dim=10))
bbranch2_node = bbranch2(bx2)
bz = BLayer.merge([bbranch1_node, bbranch2_node], mode="concat")
bmodel = BModel([bx1_1, bx2], bz)
kx1 = KLayer.Input(shape=(10, ))
kx2 = KLayer.Input(shape=(10, ))
ky1 = KLayer.Dense(12, activation="sigmoid")(kx1)
kbranch1_node = KModel(kx1, ky1)(kx1)
kbranch2 = KSequential()
kbranch2.add(KLayer.Dense(12, input_dim=10))
kbranch2_node = kbranch2(kx2)
kz = KLayer.merge([kbranch1_node, kbranch2_node], mode="concat")
kmodel = KModel([kx1, kx2], kz)
input_data = [np.random.random([2, 10]), np.random.random([2, 10])]
self.compare_newapi(kmodel, bmodel, input_data,
self.convert_two_dense_model)
def test_graph(self):
x1 = BLayer.Input(input_shape=(8, ))
x2 = BLayer.Input(input_shape=(6, ))
y1 = BLayer.Dense(10)(x1)
y2 = BLayer.Dense(10)(x2)
model = BLayer.Model([x1, x2], [y1, y2])
input_shapes = model.get_input_shape()
np.testing.assert_allclose((8, ), input_shapes[0][1:])
np.testing.assert_allclose((6, ), input_shapes[1][1:])
def test_train_dataset(self):
images = []
labels = []
for i in range(0, 8):
features = np.random.uniform(0, 1, (200, 200, 3))
label = np.array([2])
images.append(features)
labels.append(label)
image_frame = DistributedImageFrame(self.sc.parallelize(images),
self.sc.parallelize(labels))
transformer = Pipeline([
BytesToMat(),
Resize(256, 256),
CenterCrop(224, 224),
ChannelNormalize(0.485, 0.456, 0.406, 0.229, 0.224, 0.225),
MatToTensor(),
ImageFrameToSample(target_keys=['label'])
])
data_set = DataSet.image_frame(image_frame).transform(transformer)
model = BSequential()
model.add(BLayer.Convolution2D(1, 5, 5, input_shape=(3, 224, 224)))
model.add(BLayer.Reshape((1 * 220 * 220, )))
model.add(BLayer.Dense(20, activation="softmax"))
model.compile(optimizer="sgd",
loss="sparse_categorical_crossentropy",
metrics=["accuracy"])
model.fit(data_set, batch_size=8, nb_epoch=2, validation_data=data_set)
def test_merge_method_sum(self):
bx1 = BLayer.Input(shape=(8, ))
bx2 = BLayer.Input(shape=(6, ))
by1 = BLayer.Dense(10)(bx1)
by2 = BLayer.Dense(10)(bx2)
bz = BLayer.merge([by1, by2], mode="sum")
bmodel = BModel([bx1, bx2], bz, name="graph1")
kx1 = KLayer.Input(shape=(8, ))
kx2 = KLayer.Input(shape=(6, ))
ky1 = KLayer.Dense(10)(kx1)
ky2 = KLayer.Dense(10)(kx2)
kz = kmerge([ky1, ky2], mode="sum")
kmodel = KModel([kx1, kx2], kz)
input_data = [np.random.random([2, 8]), np.random.random([2, 6])]
self.compare_newapi(kmodel, bmodel, input_data,
self.convert_two_dense_model)
def test_merge_method_model_concat(self):
bx1 = BLayer.Input(shape=(4, ))
bx2 = BLayer.Input(shape=(5, ))
by1 = BLayer.Dense(6, activation="sigmoid")(bx1)
bbranch1 = BModel(bx1, by1)(bx1)
bbranch2 = BLayer.Dense(8)(bx2)
bz = BLayer.merge([bbranch1, bbranch2], mode="concat")
bmodel = BModel([bx1, bx2], bz)
kx1 = KLayer.Input(shape=(4, ))
kx2 = KLayer.Input(shape=(5, ))
ky1 = KLayer.Dense(6, activation="sigmoid")(kx1)
kbranch1 = KModel(kx1, ky1)(kx1)
kbranch2 = KLayer.Dense(8)(kx2)
kz = KLayer.merge([kbranch1, kbranch2], mode="concat")
kmodel = KModel([kx1, kx2], kz)
input_data = [np.random.random([2, 4]), np.random.random([2, 5])]
self.compare_newapi(kmodel, bmodel, input_data,
self.convert_two_dense_model)
def test_train(self):
x = np.random.random([32, 10])
y = np.random.random([
32,
])
model = BSequential()
model.add(BLayer.Dense(5, input_shape=(10, )))
model.compile(optimizer="sgd", loss="mse", metrics=["accuracy"])
model.fit(x, y, batch_size=8, nb_epoch=2, validation_data=(x, y))
model.evaluate(x, y, batch_size=8)
model.predict(x)
