#
# SPDX-FileCopyrightText: Copyright (c) 2021-2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
# SPDX-License-Identifier: Apache-2.0
#
"""Layers implementing downsampling"""
from tensorflow.keras.layers import Layer
from tensorflow.experimental.numpy import swapaxes
[docs]class Downsampling(Layer):
# pylint: disable=line-too-long
"""Downsampling(samples_per_symbol, offset=0, num_symbols=None, axis=-1, **kwargs)
Downsamples a tensor along a specified axis by retaining one out of
``samples_per_symbol`` elements.
Parameters
----------
samples_per_symbol: int
The downsampling factor. If ``samples_per_symbol`` is equal to `n`, then the
downsampled axis will be `n`-times shorter.
offset: int
Defines the index of the first element to be retained.
Defaults to zero.
num_symbols: int
Defines the total number of symbols to be retained after
downsampling.
Defaults to None (i.e., the maximum possible number).
axis: int
The dimension to be downsampled. Must not be the first dimension.
Input
-----
x : [...,n,...], tf.DType
The tensor to be downsampled. `n` is the size of the `axis` dimension.
Output
------
y : [...,k,...], same dtype as ``x``
The downsampled tensor, where ``k``
is min((``n``-``offset``)//``samples_per_symbol``, ``num_symbols``).
"""
def __init__(self,
samples_per_symbol,
offset=0,
num_symbols=None,
axis=-1, **kwargs):
super().__init__(**kwargs)
self._samples_per_symbol = samples_per_symbol
self._offset = offset
self._num_symbols = num_symbols
self._axis = axis
def call(self, inputs):
# Put selected axis last
x = swapaxes(inputs, self._axis, -1)
# Downsample
x = x[...,self._offset::self._samples_per_symbol]
if self._num_symbols is not None:
x = x[...,:self._num_symbols]
# Put last axis to original position
x = swapaxes(x, -1, self._axis)
return x