如何在Tensorflow或Keras中实现K-Max池化? [英] How to implement K-Max pooling in Tensorflow or Keras?

问题描述

首先，我知道我应该使用top_k，但是k-max池很难实现(在TF中实现)的原因是必须保留顺序.

First off I know that I should use top_k but what makes k-max pooling hard (to implement in TF) is that it has to preserve the order.

我到目前为止有什么:

import tensorflow as tf
from tensorflow.contrib.framework import sort
sess = tf.Session()
a = tf.convert_to_tensor([[[5, 1, 10, 2], [3, 11, 2, 6]]])
b = sort(tf.nn.top_k(a, k=2)[1])
print(tf.gather(a, b, axis=-1).eval(session=sess))

已经很近了，但是还没到

it's close but not there yet

我得到的是:[[[[[[5，10]，[1，2]]]，[[[3，2]，[11，6]]]]]

what I get: [[[[[ 5, 10], [ 1, 2]]], [[[ 3, 2], [11, 6]]]]]

我想要什么:[[[5，10]，[11，6]]

what I want: [[[5, 10], [11, 6]]]

我几乎百分百地确定是否需要collect_nd，但我无法弄清楚，我也是pytorch用户，在那真的很容易

I am almost hundred percent sure that gather_nd is required but I can't figure that out, also I am a pytorch user and it's really easy there

import torch
a = torch.LongTensor([[[5, 1, 10, 2], [3, 11, 2, 6]]])
b = a.topk(2, dim = -1)[1].sort(dim = -1)[0]
print(a.gather(-1, b))

哦，还有我发现的每个代码都不是一个保存命令的命令(从语义上来说是错误的)

Oh and also every code that I found was not an order preserving one(which is semantically wrong)

推荐答案

奇怪...这应该超级简单，但我们找不到合适的解决方案...

Weird... this should be super easy and yet we can't find a ready solution...

尝试一下:

sess = tf.Session()
k = 2
a = tf.convert_to_tensor([[[5, 1, 10, 2], [3, 11, 2, 6]]])
b = tf.nn.top_k(a, k=k, sorted=True)[1]
b = sort(b)

flatA = tf.reshape(a,(-1,))
shapeA = tf.shape(a)
lenA = tf.shape(flatA)[0]
kShape = tf.concat([shapeA[:-1],tf.constant([k])], axis=-1)

indices = tf.range(lenA)
indices = tf.reshape(indices,shapeA)

toSum = tf.expand_dims(tf.gather(indices,0,axis=-1), axis=-1)
b += toSum
b = tf.reshape(b,(-1,))

gat = tf.gather(flatA, b)
gat = tf.reshape(gat, kShape)

print(gat.eval(session=sess))

这篇关于如何在Tensorflow或Keras中实现K-Max池化?的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！