You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
DescEx="<p>Returns the logarithm of the sum of the exponentials of the given array of floating pointing point numbers.</p><p>An <var>EArgumentException</var> exception is raised if the array is empty.</p>"
1848
+
Extra="<p>The mathematical definition of <em>LSE</em> is <em>LSE(x1,...,xn) = log(exp(x1) + ... + exp(xn))</em>. The version of the algorithm used here uses an <a href="https://gregorygundersen.com/blog/2020/02/09/log-sum-exp/">algebraic trick</a> to minimise the risk of overflow.</p><p>For more information see <a href="https://nhigham.com/2021/01/05/what-is-the-log-sum-exp-function/">What Is the Log-Sum-Exp Function?</a> by Nick Higham.</p>"
DescEx="<p>Applies the <a href="https://en.wikipedia.org/wiki/Softmax_function"><em>softmax</em></a> function to each element of floating point array <var>A</var> and normalizes them into a probability distribution proportional to the exponentials of the elements of <var>A</var>. The normalised values are returned as an array of the same size as <var>A</var>.</p><p>An <var>EArgumentException</var> exception is raised if <var>A</var> is empty.</p>"
1863
+
Extra="<p>The <em>softmax</em> function is often used in statistics and in machine learning. It is closely related to the <em>log-sum-exp</em> (aka <em>LSE</em>) function since, for vector <strong>x</strong>=(x1,...,xn), <em>softmax(<strong>x</strong>)=exp[<strong>x</strong>−LSE(<strong>x</strong>)]</em>.</p>"
0 commit comments