Tensorflow multiply 3D batch tensor with a 2D weight - tensorflow

I've got two tensors with the shape shown below,
batch.shape = [?, 5, 4]
weight.shape = [3, 5]
by multiplying the weight with every element in the batch, I want to get
result.shape = [?, 3, 4]
what is the most efficient way to achieve this?

Try this:
newbatch = tf.transpose(batch,[1,0,2])
newbatch = tf.reshape(newbatch,[5,-1])
result = tf.matmul(weight,newbatch)
result = tf.reshape(result,[3,-1,4])
result = tf.transpose(result, [1,0,2])
Or more compactly:
newbatch = tf.reshape(tf.transpose(batch,[1,0,2]),[5,-1])
result = tf.transpose(tf.reshape(tf.matmul(weight,newbatch),[3,-1,4]), [1,0,2])

Try this:
tf.einsum("ijk,aj-> iak",batch,weight)
A generalized contraction between tensors of arbitrary dimension Refer this for more information

Related

How to concatenate two tensors with intervals in tensorflow?

I want to concatenate two tensors checkerboard-ly in tensorflow2, like examples showed below:
example 1:
a = [[1,1],[1,1]]
b = [[0,0],[0,0]]
concated_a_and_b = [[1,0,1,0],[0,1,0,1]]
example 2:
a = [[1,1,1],[1,1,1],[1,1,1]]
b = [[0,0,0],[0,0,0],[0,0,0]]
concated_a_and_b = [[1,0,1,0,1,0],[0,1,0,1,0,1],[1,0,1,0,1,0]]
Is there a decent way in tensorflow2 to concatenate them like this?
A bit of background for this:
I first split a tensor c with a checkerboard mask into two halves a and b. A after some transformation I have to concat them back into oringnal shape and order.
What I mean by checkerboard-ly:
Step 1: Generate a matrix with alternated values
You can do this by first concatenating into [1, 0] pairs, and then by applying a final reshape.
Step 2: Reverse some rows
I split the matrix into two parts, reverse the second part and then rebuild the full matrix by picking alternatively from the first and second part
Code sample:
import math
import numpy as np
import tensorflow as tf
a = tf.ones(shape=(3, 4))
b = tf.zeros(shape=(3, 4))
x = tf.expand_dims(a, axis=-1)
y = tf.expand_dims(b, axis=-1)
paired_ones_zeros = tf.concat([x, y], axis=-1)
alternated_values = tf.reshape(paired_ones_zeros, [-1, a.shape[1] + b.shape[1]])
num_samples = alternated_values.shape[0]
middle = math.ceil(num_samples / 2)
is_num_samples_odd = middle * 2 != num_samples
# Gather first part of the matrix, don't do anything to it
first_elements = tf.gather_nd(alternated_values, [[index] for index in range(middle)])
# Gather second part of the matrix and reverse its elements
second_elements = tf.reverse(tf.gather_nd(alternated_values, [[index] for index in range(middle, num_samples)]), axis=[1])
# Pick alternatively between first and second part of the matrix
indices = np.concatenate([[[index], [index + middle]] for index in range(middle)], axis=0)
if is_num_samples_odd:
indices = indices[:-1]
output = tf.gather_nd(
tf.concat([first_elements, second_elements], axis=0),
indices
)
print(output)
I know this is not a decent way as it will affect time and space complexity. But it solves the above problem
def concat(tf1, tf2):
result = []
for (index, (tf_item1, tf_item2)) in enumerate(zip(tf1, tf2)):
item = []
for (subitem1, subitem2) in zip(tf_item1, tf_item2):
if index % 2 == 0:
item.append(subitem1)
item.append(subitem2)
else:
item.append(subitem2)
item.append(subitem1)
concated_a_and_b.append(item)
return concated_a_and_b

rebuild torch tensor from its batchified version

This is very nice example of how to build a 3D tensor:
import torch
y = torch.rand(100, 1)
batch_size = 10
batched_data = y.contiguous().view(batch_size, -1, y.size(-1)).transpose(0,1)
batched_data.shape
the output is:
torch.Size([10, 10, 1])
ok, now what I want to do is, starting from batched_data I want to build y.
The other way around.
Any good suggestion with a powerful pytorch streamlined code?
==== Additional input =====
I am using this for RNN and now I have some doubts, becaus eif you consider the following code:
import torch
y = torch.arange(100).view(100,1)
batch_size = 10
batched_data = y.contiguous().view(batch_size, -1, y.size(-1)).transpose(0,1)
batched_data.shape
The output is:
tensor([[[ 0],
[10],
[20],
[30],
[40],
[50],
[60],
[70],
[80],
[90]],
[[ 1],
[11],
[21],
[31],
[41],
[51],
[61],
[71],
[81],
[91]],
Which I would not expect. I would expect something like:
[[1,2,3,4,5,6,7,8,9,10],[11,12,13,14,15,16,17,18,19,20],....
Suppose you want to do something like this to rebuild y:
rebuilded_y = batched_data.transpose(0,1).view(*y.shape)
To make the input look like you expected you need to remove transpose and additional dimension in batched_data:
batched_data = y.contiguous().view(batch_size, -1)
If you want to prepare inputs for RNN, you need to know that RNN takes 3d tensors of shape, seq_len, batch, input_size. Here, input_size refers to the number of features and in your scenario, it is 1. So, the input tensor of shape 10, 10, 1 can still be a valid input for an RNN.
Example
rnn = nn.RNN(input_size=1, hidden_size=20, num_layers=1)
input = torch.randn(10, 10, 1)
output, hn = rnn(input)
print(output.size()) # 10, 10, 20
RNN's output is of shape, seq_len, batch, num_directions * hidden_size.

Tensorflow: When using slim.dataset.Dataset, is there a way to map label ID values to other values?

dataset = slim.dataset.Dataset(...)
provider = slim.dataset_data_provider.DatasetDataProvider(dataset, ..._
image, labels = provider.get(['image', 'label')
Let's say, for an example in a dataset A, labels could be [1, 2, 1, 3]. However, for some reason (e.g, due to dataset B), I would like to map the label IDs to other values. The mapping could be like below.
# {old_label: target_label}
mapping = {0: 0, 1: 2, 2: 2, 3: 2, 4: 2, 5: 3, 6: 1}
For now, I am guessing two ways:
-- tf.data.Dataset seems to have a map(map_func) function that every examples should pass, which could be the solution. However, I am more familiar to slim.dataset.Dataset. Is there a similar trick for slim.dataset.Dataset?
-- I was wondering if I can simply apply some mapping function to a tensor label such as:
new_labels = tf.map_fn(lambda x: x+1, labels, dtype=tf.int32)
# labels = [1 2 1 3] --> new_labels = [2 3 2 4]. This works.
new_labels = tf.map_fn(lambda x: mapping[x], labels, dtype=tf.int32)
# I wished but this does not work!
However, the below didn't work, which is what I need. Could anyone please advise?
I think you can try tf.contrib.lookup:
keys = list(mapping.keys())
values = [mapping[k] for k in keys]
table = tf.contrib.lookup.HashTable(
tf.contrib.lookup.KeyValueTensorInitializer(keys, values, key_dtype=tf.int64, value_dtype=tf.int64), -1
)
new_labels = table.lookup(labels)
sess=tf.Session()
sess.run(table.init)
print(sess.run(new_labels))

Row-wise Histogram

Given a 2-dimensional tensor t, what's the fastest way to compute a tensor h where
h[i, :] = tf.histogram_fixed_width(t[i, :], vals, nbins)
I.e. where tf.histogram_fixed_width is called per row of the input tensor t?
It seems that tf.histogram_fixed_width is missing an axis parameter that works like, e.g., tf.reduce_sum's axis parameter.
tf.histogram_fixed_width works on the entire tensor indeed. You have to loop through the rows explicitly to compute the per-row histograms. Here is a complete working example using TensorFlow's tf.while_loop construct :
import tensorflow as tf
t = tf.random_uniform([2, 2])
i = 0
hist = tf.constant(0, shape=[0, 5], dtype=tf.int32)
def loop_body(i, hist):
h = tf.histogram_fixed_width(t[i, :], [0.0, 1.0], nbins=5)
return i+1, tf.concat_v2([hist, tf.expand_dims(h, 0)], axis=0)
i, hist = tf.while_loop(
lambda i, _: i < 2, loop_body, [i, hist],
shape_invariants=[tf.TensorShape([]), tf.TensorShape([None, 5])])
sess = tf.InteractiveSession()
print(hist.eval())
Inspired by keveman's answer and because the number of rows of t is fixed and rather small, I chose to use a combination of tf.gather to split rows and tf.pack to join rows. It looks simple and works, will see if it is efficient...
t_histo_rows = [
tf.histogram_fixed_width(
tf.gather(t, [row]),
vals, nbins)
for row in range(t_num_rows)]
t_histo = tf.pack(t_histo_rows, axis=0)
I would like to propose another implementation.
This implementation can also handle multi axes and unknown dimensions (batching).
def histogram(tensor, nbins=10, axis=None):
value_range = [tf.reduce_min(tensor), tf.reduce_max(tensor)]
if axis is None:
return tf.histogram_fixed_width(tensor, value_range, nbins=nbins)
else:
if not hasattr(axis, "__len__"):
axis = [axis]
other_axis = [x for x in range(0, len(tensor.shape)) if x not in axis]
swap = tf.transpose(tensor, [*other_axis, *axis])
flat = tf.reshape(swap, [-1, *np.take(tensor.shape.as_list(), axis)])
count = tf.map_fn(lambda x: tf.histogram_fixed_width(x, value_range, nbins=nbins), flat, dtype=(tf.int32))
return tf.reshape(count, [*np.take([-1 if a is None else a for a in tensor.shape.as_list()], other_axis), nbins])
The only slow part here is tf.map_fn but it is still faster than the other solutions mentioned.
If someone knows a even faster implementation please comment since this operation is still very expensive.
answers above is still slow running in GPU. Here i give an another option, which is faster(at least in my running envirment), but it is limited to 0~1 (you can normalize the value first). the train_equal_mask_nbin can be defined once in advance
def histogram_v3_nomask(tensor, nbins, row_num, col_num):
#init mask
equal_mask_list = []
for i in range(nbins):
equal_mask_list.append(tf.ones([row_num, col_num], dtype=tf.int32) * i)
#[nbins, row, col]
#[0, row, col] is tensor of shape [row, col] with all value 0
#[1, row, col] is tensor of shape [row, col] with all value 1
#....
train_equal_mask_nbin = tf.stack(equal_mask_list, axis=0)
#[inst, doc_len] float to int(equaly seg float in bins)
int_input = tf.cast(tensor * (nbins), dtype=tf.int32)
#input [row,col] -> copy N times, [nbins, row_num, col_num]
int_input_nbin_copy = tf.reshape(tf.tile(int_input, [nbins, 1]), [nbins, row_num, col_num])
#calculate histogram
histogram = tf.transpose(tf.count_nonzero(tf.equal(train_equal_mask_nbin, int_input_nbin_copy), axis=2))
return histogram
With the advent of tf.math.bincount, I believe the problem has become much simpler.
Something like this should work:
def hist_fixed_width(x,st,en,nbins):
x=(x-st)/(en-st)
x=tf.cast(x*nbins,dtype=tf.int32)
x=tf.clip_by_value(x,0,nbins-1)
return tf.math.bincount(x,minlength=nbins,axis=-1)

Custom median pooling in tensorflow

I am trying to implement a median pooling layer in tensorflow.
However there is neither tf.nn.median_pool and neither tf.reduce_median.
Is there a way to implement such pooling layer with the python api ?
You could use something like:
patches = tf.extract_image_patches(tensor, [1, k, k, 1], ...)
m_idx = int(k*k/2+1)
top = tf.top_k(patches, m_idx, sorted=True)
median = tf.slice(top, [0, 0, 0, m_idx-1], [-1, -1, -1, 1])
To accommodate even sized median kernels and multiple channels, you will need to extend this, but this should get you most of the way.
As of March 2017, an easier answer (that under the hood works similarly to how Alex suggested) is to do this:
patches = tf.extract_image_patches(x, [1, k, k, 1], [1, k, k, 1], 4*[1], 'VALID')
medians = tf.contrib.distributions.percentile(patches, 50, axis=3)
For me, Alex's answer is not working for tf 1.4.1.
tf.top_k should be tf.nn.top_k
and should get values of tf.nn.top_k
Also, if the input is [1, H, W, C], either answer could not only work on height and width and neglect the channel.
Channel-wise median-pooling can be done by some addition reshapes on top of other answers:
# assuming NHWC layout
strides = rates = [1, 1, 1, 1]
patches = tf.extract_image_patches(x, [1, k, k, 1], strides, rates, 'VALID')
batch_size = tf.shape(x)[0]
n_channels = tf.shape(x)[-1]
n_patches_h = (tf.shape(x)[1] - k) // strides[1] + 1
n_patches_w = (tf.shape(x)[2] - k) // strides[2] + 1
n_patches = tf.shape(patches)[-1] // n_channels
patches = tf.reshape(patches, [batch_size, k, k, n_patches_h * n_patches_w, n_channels])
medians = tf.contrib.distributions.percentile(patches, 50, axis=[1,2])
medians = tf.reshape(medians, (batch_size, n_patches_h, n_patches_w, n_channels))
Not very efficient though.
I was looking for a median filter for tensorflowjs but can't seem to find one. tfa has a median filter now I think but for tf.js you can use this. Not sure if it would work on nodegpu.
function medianFilter(x, filter, strides, pad) {
//make Kernal
//todo allow for filter as array or number
let filterSize = filter ** 2;
let locs = tf.range(0, filterSize, filterSize );
//makes a bunc of arrays each one reprensentin one of the valuesin the median window ie 2x2 filter i in chanle and 4 out chanles
let f = tf.oneHot(tf.range(0,filterSize,1, 'int32'), filterSize).reshape([filter, filter, 1, filterSize]);
let y = tf.conv2d(x,f,strides,pad);
let m_idx = Math.floor(filterSize/2)+1;
let top = tf.topk(y, m_idx, true);
//note that thse are 3d tensors and if you use 4d ones add a 0 and -1 infron like in above ansowers
let median = tf.slice(top.values, [0,0,m_idx-1], [-1,-1,1] );
return median;
}