activation_sparsemax

For each batch <code>i</code>, and class <code>j</code>,
sparsemax activation function is defined as:
<code>sparsemax(x)[i, j] = max(x[i, j] - (x[i, :]), 0).</code>

Interface to 'Keras' <https://keras.io>, a high-level neural
networks API. 'Keras' was developed with a focus on enabling fast experimentation,
supports both convolution based networks and recurrent networks (as well as
combinations of the two), and runs seamlessly on both CPU and GPU devices.

Tomasz Kalinowski

keras3

R Interface to 'Keras'

Daniel Falbel

JJ Allaire

François Chollet

Posit Software, PBC 

Google 

Yuan Tang

Wouter Van Der Bijl

Martin Studer

Sigrid Keydana

activation_sparsemax function

<dl><dt>x</dt>
<dd>Input tensor.</dd>
<dt>axis</dt>
<dd><code>int</code>, axis along which the sparsemax operation is applied. (1-based)</dd></dl>

Arguments

<ul>
<li><a href="https://arxiv.org/abs/1602.02068">Martins et.al., 2016</a></li>
</ul>

Reference

Sparsemax activation function. — activation_sparsemax

<dl>

<dt>x</dt>
<dd>Input tensor.</dd>


<dt>axis</dt>
<dd><code>int</code>, axis along which the sparsemax operation is applied. (1-based)</dd>

</dl>

<ul>
<li><a href='https://arxiv.org/abs/1602.02068'>Martins et.al., 2016</a></li>
</ul>

activation_sparsemax: Sparsemax activation function.

Description

Usage

Value

Arguments

Reference

See Also