有扩展现有激活功能的简单方法吗?我的自定义softmax函数返回:一个操作的梯度没有“无”

时间:2019-03-08 16:31:34

标签: python tensorflow keras softmax activation-function

我想尝试通过仅使用向量中的前k个值来使softmax更快。

为此,我尝试为tensorflow实现自定义函数以在模型中使用:

def softmax_top_k(logits, k=10):
    values, indices = tf.nn.top_k(logits, k, sorted=False)
    softmax = tf.nn.softmax(values)
    logits_shape = tf.shape(logits)
    return_value = tf.sparse_to_dense(indices, logits_shape, softmax)
    return_value = tf.convert_to_tensor(return_value, dtype=logits.dtype, name=logits.name)
    return return_value

我正在使用mnist进行测试,以了解该尝试是否有效:

fashion_mnist = keras.datasets.fashion_mnist
(train_images, train_labels), (test_images, test_labels) = fashion_mnist.load_data()

# normalize the data
train_images = train_images / 255.0
test_images = test_images / 255.0

# split the training data into train and validate arrays (will be used later)
train_images, train_images_validate, train_labels, train_labels_validate = train_test_split(
    train_images, train_labels, test_size=0.2, random_state=133742,
)

model = keras.models.Sequential([
    keras.layers.Flatten(input_shape=(28, 28)),
    keras.layers.Dense(128, activation=tf.nn.relu),
    keras.layers.Dense(10, activation=softmax_top_k)
])


model.compile(
    loss='sparse_categorical_crossentropy',
    optimizer='adam',
    metrics=['accuracy']
)

model.fit(
    train_images, train_labels,
    epochs=10,
    validation_data=(train_images_validate, train_labels_validate),
)

model_without_cnn.compile(
    loss='sparse_categorical_crossentropy',
    optimizer='adam',
    metrics=['accuracy']
)

model_without_cnn.fit(
    train_images, train_labels,
    epochs=10,
    validation_data=(train_images_validate, train_labels_validate),
)

但是在执行过程中发生了错误:

ValueError: An operation hasfor gradient. Please make sure that all of your ops have a gradient defined (i.e. are differentiable).

我发现了this: (How to make a custom activation function),它解释了如何对tensorflow实现完全自定义的激活功能。但是,由于使用并扩展了softmax,我认为渐变应该仍然相同。

这是我使用python和tensorflow进行编码的第一周,因此,我还没有全面了解所有内部实现。

有没有更简单的方法可以将softmax扩展到新功能中,而不是从头开始实现呢?

谢谢!

1 个答案:

答案 0 :(得分:0)

不要使用稀疏张量来使张量具有“除了softmaxed top-K值之外的所有零”,而应使用tf.scatter_nd

import tensorflow as tf

def softmax_top_k(logits, k=10):
    values, indices = tf.nn.top_k(logits, k, sorted=False)
    softmax = tf.nn.softmax(values)
    logits_shape = tf.shape(logits)
    # Assuming that logits is 2D
    rows = tf.tile(tf.expand_dims(tf.range(logits_shape[0]), 1), [1, k])
    scatter_idx = tf.stack([rows, indices], axis=-1)
    return tf.scatter_nd(scatter_idx, softmax, logits_shape)

编辑:这是具有任意数量尺寸的张量的稍微复杂一些的版本。但是,该代码仍然要求在图构建时知道维数。

import tensorflow as tf

def softmax_top_k(logits, k=10):
    values, indices = tf.nn.top_k(logits, k, sorted=False)
    softmax = tf.nn.softmax(values)
    # Make nd indices
    logits_shape = tf.shape(logits)
    dims = [tf.range(logits_shape[i]) for i in range(logits_shape.shape.num_elements() - 1)]
    grid = tf.meshgrid(*dims, tf.range(k), indexing='ij')
    scatter_idx = tf.stack(grid[:-1] + [indices], axis=-1)
    return tf.scatter_nd(scatter_idx, softmax, logits_shape)