Python layers.Permute方法代码示例

# 需要导入模块: from tensorflow.keras import layers [as 别名]
# 或者: from tensorflow.keras.layers import Permute [as 别名]
def call(self, inputs):  # (B, S, H)
        # Expand weights to include batch size through implicit broadcasting
        W1, W2 = self.W1[None, :, :], self.W2[None, :, :]
        hidden_states_transposed = Permute(dims=(2, 1))(inputs)                                     # (B, H, S)
        attention_score = tf.matmul(W1, hidden_states_transposed)                                   # (B, size, S)
        attention_score = Activation('tanh')(attention_score)                                       # (B, size, S)
        attention_weights = tf.matmul(W2, attention_score)                                          # (B, num_hops, S)
        attention_weights = Activation('softmax')(attention_weights)                                # (B, num_hops, S)
        embedding_matrix = tf.matmul(attention_weights, inputs)                                     # (B, num_hops, H)
        embedding_matrix_flattened = Flatten()(embedding_matrix)                                    # (B, num_hops*H)

        if self.use_penalization:
            attention_weights_transposed = Permute(dims=(2, 1))(attention_weights)                  # (B, S, num_hops)
            product = tf.matmul(attention_weights, attention_weights_transposed)                    # (B, num_hops, num_hops)
            identity = tf.eye(self.num_hops, batch_shape=(inputs.shape[0],))                        # (B, num_hops, num_hops)
            frobenius_norm = tf.sqrt(tf.reduce_sum(tf.square(product - identity)))  # distance
            self.add_loss(self.penalty_coefficient * frobenius_norm)  # loss

        if self.model_api == 'functional':
            return embedding_matrix_flattened, attention_weights
        elif self.model_api == 'sequential':
            return embedding_matrix_flattened 

# 需要导入模块: from tensorflow.keras import layers [as 别名]
# 或者: from tensorflow.keras.layers import Permute [as 别名]
def squeeze_excite_block(input_tensor, ratio=16):
    """ Create a channel-wise squeeze-excite block

    Args:
        input_tensor: input Keras tensor
        ratio: number of output filters

    Returns: a Keras tensor

    References
    -   [Squeeze and Excitation Networks](https://arxiv.org/abs/1709.01507)
    """
    init = input_tensor
    channel_axis = 1 if K.image_data_format() == "channels_first" else -1
    filters = _tensor_shape(init)[channel_axis]
    se_shape = (1, 1, filters)

    se = GlobalAveragePooling2D()(init)
    se = Reshape(se_shape)(se)
    se = Dense(filters // ratio, activation='relu', kernel_initializer='he_normal', use_bias=False)(se)
    se = Dense(filters, activation='sigmoid', kernel_initializer='he_normal', use_bias=False)(se)

    if K.image_data_format() == 'channels_first':
        se = Permute((3, 1, 2))(se)

    x = multiply([init, se])
    return x 

# 需要导入模块: from tensorflow.keras import layers [as 别名]
# 或者: from tensorflow.keras.layers import Permute [as 别名]
def channel_squeeze_excite_block(input, ratio=0.25):
    init = input
    channel_axis = 1 if K.image_data_format() == "channels_first" else -1
    filters = init._keras_shape[channel_axis]
    cse_shape = (1, 1, filters)

    cse = layers.GlobalAveragePooling2D()(init)
    cse = layers.Reshape(cse_shape)(cse)
    ratio_filters = int(np.round(filters * ratio))
    if ratio_filters < 1:
        ratio_filters += 1
    cse = layers.Conv2D(
        ratio_filters,
        (1, 1),
        padding="same",
        activation="relu",
        kernel_initializer="he_normal",
        use_bias=False,
    )(cse)
    cse = layers.BatchNormalization()(cse)
    cse = layers.Conv2D(
        filters,
        (1, 1),
        activation="sigmoid",
        kernel_initializer="he_normal",
        use_bias=False,
    )(cse)

    if K.image_data_format() == "channels_first":
        cse = layers.Permute((3, 1, 2))(cse)

    cse = layers.Multiply()([init, cse])
    return cse 

# 需要导入模块: from tensorflow.keras import layers [as 别名]
# 或者: from tensorflow.keras.layers import Permute [as 别名]
def squeeze_excite_block(input, ratio=16):
    ''' Create a channel-wise squeeze-excite block

    Args:
        input: input tensor
        filters: number of output filters

    Returns: a keras tensor

    References
    -   [Squeeze and Excitation Networks](https://arxiv.org/abs/1709.01507)
    '''
    init = input
    channel_axis = 1 if K.image_data_format() == "channels_first" else -1
    filters = init._keras_shape[channel_axis]
    se_shape = (1, 1, filters)

    se = GlobalAveragePooling2D()(init)
    se = Reshape(se_shape)(se)
    se = Dense(filters // ratio, activation='relu', kernel_initializer='he_normal', use_bias=False)(se)
    se = Dense(filters, activation='sigmoid', kernel_initializer='he_normal', use_bias=False)(se)

    if K.image_data_format() == 'channels_first':
        se = Permute((3, 1, 2))(se)

    x = multiply([init, se])
    return x 

# 需要导入模块: from tensorflow.keras import layers [as 别名]
# 或者: from tensorflow.keras.layers import Permute [as 别名]
def build_and_load_model(model_capacity):
    """
    Build the CNN model and load the weights

    Parameters
    ----------
    model_capacity : 'tiny', 'small', 'medium', 'large', or 'full'
        String specifying the model capacity, which determines the model's
        capacity multiplier to 4 (tiny), 8 (small), 16 (medium), 24 (large),
        or 32 (full). 'full' uses the model size specified in the paper,
        and the others use a reduced number of filters in each convolutional
        layer, resulting in a smaller model that is faster to evaluate at the
        cost of slightly reduced pitch estimation accuracy.

    Returns
    -------
    model : tensorflow.keras.models.Model
        The pre-trained keras model loaded in memory
    """
    from tensorflow.keras.layers import Input, Reshape, Conv2D, BatchNormalization
    from tensorflow.keras.layers import MaxPool2D, Dropout, Permute, Flatten, Dense
    from tensorflow.keras.models import Model

    if models[model_capacity] is None:
        capacity_multiplier = {
            'tiny': 4, 'small': 8, 'medium': 16, 'large': 24, 'full': 32
        }[model_capacity]

        layers = [1, 2, 3, 4, 5, 6]
        filters = [n * capacity_multiplier for n in [32, 4, 4, 4, 8, 16]]
        widths = [512, 64, 64, 64, 64, 64]
        strides = [(4, 1), (1, 1), (1, 1), (1, 1), (1, 1), (1, 1)]

        x = Input(shape=(1024,), name='input', dtype='float32')
        y = Reshape(target_shape=(1024, 1, 1), name='input-reshape')(x)

        for l, f, w, s in zip(layers, filters, widths, strides):
            y = Conv2D(f, (w, 1), strides=s, padding='same',
                       activation='relu', name="conv%d" % l)(y)
            y = BatchNormalization(name="conv%d-BN" % l)(y)
            y = MaxPool2D(pool_size=(2, 1), strides=None, padding='valid',
                          name="conv%d-maxpool" % l)(y)
            y = Dropout(0.25, name="conv%d-dropout" % l)(y)

        y = Permute((2, 1, 3), name="transpose")(y)
        y = Flatten(name="flatten")(y)
        y = Dense(360, activation='sigmoid', name="classifier")(y)

        model = Model(inputs=x, outputs=y)

        package_dir = os.path.dirname(os.path.realpath(__file__))
        filename = "model-{}.h5".format(model_capacity)
        model.load_weights(os.path.join(package_dir, filename))
        model.compile('adam', 'binary_crossentropy')

        models[model_capacity] = model

    return models[model_capacity]