<html><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8"></head><body dir="auto"><div dir="auto">Lundberg and Lee proposed to use Shapley values as a general method to explain all kinds of machine learning models. As I understand it this works by varying input values systematically and then observing how much the model output changes. </div><div dir="auto">https://arxiv.org/abs/1705.07874v2</div><div dir="auto"><br></div><div dir="auto">It is one of the most common methods used today to explain neural networks. Here is a recent paper from researchers at MIT and Microsoft Research about the topic:</div><div dir="auto">https://arxiv.org/abs/2104.14403</div><div dir="auto"><br></div><div dir="auto">The training set in my case was just an Excel/CSV file where I estimated the feature values myself. This means I have asked myself for reach country and each feature which value would describe the situation best on a scale from -1 to 1.</div><div dir="auto"><br></div><div dir="auto">I think it is an interesting approach. Maybe I can improve the data or the model a bit and make a blog post out of it.</div><div dir="auto"><br></div><div dir="auto">-J.</div><div dir="auto"><br></div><div><br></div><div align="left" dir="auto" style="font-size:100%;color:#000000"><div>-------- Original message --------</div><div>From: Steve Smith <sasmyth@swcp.com> </div><div>Date: 7/15/22 10:29 PM (GMT+01:00) </div><div>To: friam@redfish.com </div><div>Subject: Re: [FRIAM] Modeling Fascism </div><div><br></div></div>
<p>for those of us trying to suss out the implications of this, I
offer:</p>
<p> <a href="https://en.wikipedia.org/wiki/Shapley_value" class="moz-txt-link-freetext">https://en.wikipedia.org/wiki/Shapley_value</a></p>
<p>I'm not clear on a number of things Jochen:</p>
<ol>
<li>The Shapley values implies an iterated/cooperative model?</li>
<li>I see how the "training set" from post WWI Europe might be
used in this context.</li>
<li>How do your assumptions about Trump-era America fit in? <br>
</li>
<ol>
<li>Is the assumption that post WWI (hidden) conditions in
Europe are similar enough to contemporary Global conditions?</li>
</ol>
</ol>
<p>I'm probably missing something (a lot)?</p>
<p>- Steve<br>
</p>
<div class="moz-cite-prefix">On 7/15/22 2:06 PM, Jochen Fromm wrote:<br>
</div>
<blockquote type="cite">
<div dir="auto">Today after work I've tried to built a small and
simple machine learning model to predict fascism, based on 6 or
7 fundamental features. Using Shapley values we can see which
feature contributes the most to the outbreak of fascism. If I
enter values which fit to the US under Trump the model indeed
predicts a form of authoritarianism.</div>
<div dir="auto"><br>
</div>
<div dir="auto">No big data or deep learning, just a small neural
network based on scikit-learn (no Tensorflow, Keras, or
Pytorch). Juypter notebook here:</div>
<div dir="auto"><a href="https://nbviewer.org/github/JochenFromm/JupyterNotebooks/blob/master/ModelingFascism.ipynb" class="moz-txt-link-freetext">https://nbviewer.org/github/JochenFromm/JupyterNotebooks/blob/master/ModelingFascism.ipynb</a></div>
<div dir="auto"><br>
</div>
<div dir="auto">-J.</div>
<div dir="auto"><br>
</div>
<br>
<fieldset class="moz-mime-attachment-header"></fieldset>
<pre wrap="" class="moz-quote-pre">-. --- - / ...- .- .-.. .. -.. / -- --- .-. ... . / -.-. --- -.. .
FRIAM Applied Complexity Group listserv
Fridays 9a-12p Friday St. Johns Cafe / Thursdays 9a-12p Zoom <a href="https://bit.ly/virtualfriam" class="moz-txt-link-freetext">https://bit.ly/virtualfriam</a>
to (un)subscribe <a href="http://redfish.com/mailman/listinfo/friam_redfish.com" class="moz-txt-link-freetext">http://redfish.com/mailman/listinfo/friam_redfish.com</a>
FRIAM-COMIC <a href="http://friam-comic.blogspot.com/" class="moz-txt-link-freetext">http://friam-comic.blogspot.com/</a>
archives: 5/2017 thru present <a href="https://redfish.com/pipermail/friam_redfish.com/" class="moz-txt-link-freetext">https://redfish.com/pipermail/friam_redfish.com/</a>
1/2003 thru 6/2021 <a href="http://friam.383.s1.nabble.com/" class="moz-txt-link-freetext">http://friam.383.s1.nabble.com/</a>
</pre>
</blockquote>
</body></html>