Gregor Kasieczka<p>📢📢📢 New <a href="https://sciencemastodon.com/tags/Paper" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>Paper</span></a>: '<a href="https://sciencemastodon.com/tags/FeatureSelection" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>FeatureSelection</span></a> with Distance Correlation' (<a href="https://arxiv.org/abs/2212.00046" rel="nofollow noopener noreferrer" target="_blank"><span class="invisible">https://</span><span class="">arxiv.org/abs/2212.00046</span><span class="invisible"></span></a>) - a short <a href="https://sciencemastodon.com/tags/PaperSummary" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>PaperSummary</span></a> thread </p><p>We investigates how to automatically find a small # of features that - when put into a simple <a href="https://sciencemastodon.com/tags/NeuralNetwork" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>NeuralNetwork</span></a> - yield good performance (e.g. for classification)</p><p>Two possible uses:<br>- Explain the behavior of a <a href="https://sciencemastodon.com/tags/BlackBox" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>BlackBox</span></a> classifier <br>- Build a light-weight classifier from scratch</p>