python - 使用 Lambda 创建自定义 Keras Layer 对象

Question

我想构建一个自定义 Keras 层，保留 k 个最高激活值。我目前正在这样做(并且工作正常):

def max_topk_pool(x,k):
    import tensorflow as tf
    k_max =  tf.nn.top_k(x,k=k,sorted=True,name=None)
    return  k_max

def KMax(k):
    return Lambda(max_topk_pool,
                  arguments={'k':k},
                  output_shape=lambda x: (None, k))

你知道是否有一种方法可以按照Keras在https://keras.io/layers/writing-your-own-keras-layers/中所示的方式构建自定义图层类“KMax”

from keras import backend as K
from keras.layers import Layer

class MyLayer(Layer):

def __init__(self, output_dim, **kwargs):
    self.output_dim = output_dim
    super(MyLayer, self).__init__(**kwargs)

def build(self, input_shape):
    # Create a trainable weight variable for this layer.
    self.kernel = self.add_weight(name='kernel', 
                                  shape=(input_shape[1], self.output_dim),
                                  initializer='uniform',
                                  trainable=True)
    super(MyLayer, self).build(input_shape)  # Be sure to call this at the end

def call(self, x):
    return K.dot(x, self.kernel)

def compute_output_shape(self, input_shape):
    return (input_shape[0], self.output_dim)

我想要这样的东西:

from keras import backend as K
from keras.layers import Layer

class KMax(Layer):

def __init__(self, output_dim, **kwargs):
    self.K = K
    super(MyLayer, self).__init__(**kwargs)

def build(self, input_shape):
    <... Lambda here ?>

def compute_output_shape(self, input_shape):
    return (input_shape[0], self.K)

非常感谢!

Answer 1

这是您需要的(基于 https://github.com/keras-team/keras/issues/373 ):

from keras.engine import Layer, InputSpec
from keras.layers import Flatten
import tensorflow as tf


# https://github.com/keras-team/keras/issues/373
class KMaxPooling(Layer):
    """
    K-max pooling layer that extracts the k-highest activations from a sequence (2nd dimension).
    TensorFlow backend.
    """

    def __init__(self, k=1, **kwargs):
        super().__init__(**kwargs)
        self.input_spec = InputSpec(ndim=3)
        self.k = k

    def compute_output_shape(self, input_shape):
        return input_shape[0], (input_shape[2] * self.k)

    def call(self, inputs):
        # swap last two dimensions since top_k will be applied along the last dimension
        shifted_input = tf.transpose(inputs, [0, 2, 1])

        # extract top_k, returns two tensors [values, indices]
        top_k = tf.nn.top_k(shifted_input, k=self.k, sorted=True, name=None)[0]

        # return flattened output
        return Flatten()(top_k)

    def get_config(self):
        config = {'k': self.k}
        base_config = super().get_config()
        return {**base_config, **config}