Built site for gh-pages

.nojekyll

@@ -1 +1 @@
-fa295913
+8ec74007

lectures/w09-l01.html

@@ -362,6 +362,7 @@ <h3 class="anchored" data-anchor-id="pre-work">Pre-work:</h3>
 <h3 class="anchored" data-anchor-id="in-class">In-Class</h3>
 <ol type="1">
 <li><a href="https://people.eecs.berkeley.edu/~angelopoulos/publications/downloads/gentle_intro_conformal_dfuq.pdf">A gentle introduction to Conformal Prediction and Distribution-free Uncertainty Quantification</a> <a href="https://www.youtube.com/watch?v=nql000Lu_iE">Video</a></li>
+<li><a href="https://colab.research.google.com/drive/1TC_BM7JaEYtBIq6yuYB5U4cJjeg71Tch">colab</a> from <a href="https://github.com/deel-ai/puncc">DEEL-PUNCC</a></li>
 </ol>
 </section>
 <section id="post-class" class="level3">
@@ -372,6 +373,8 @@ <h3 class="anchored" data-anchor-id="post-class">Post-class</h3>
 <li>[tools] <a href="https://github.com/valeman/awesome-conformal-prediction">awesome-conformal-prediction</a> - a collection Conformal Prediction resources including implementations.</li>
 <li>[tools] <a href="https://github.com/henrikbostrom/crepes">crepes</a> - Conformal Classifiers, Regressors, and Predictive Systems.</li>
 <li>[tools] <a href="https://github.com/ml-stat-Sustech/TorchCP">TorchCP</a> - a python toolbox for Conformal Prediction research in Deep Learning Models using PyTorch.</li>
+<li>[tools] <a href="https://github.com/scikit-learn-contrib/MAPIE">MAPIE</a> - a python toolbox for Conformal Prediction</li>
+<li>[tools] <a href="https://github.com/deel-ai/puncc">DEEL-PUNCC</a> - a python toolbox for Conformal Prediction from <a href="https://www.deel.ai/">DEEL.ai</a> a project for Dependable, Certifiable, Explainable AI for Critical Systems. Checkout the sister projects from DEEl on Bias <a href="https://github.com/deel-ai/influenciae">DEEL INFLUENCIAE</a>, <a href="https://github.com/deel-ai/oodeel">oodeel</a> for OOD, <a href="https://github.com/deel-ai/xplique">xplique</a> for XAI,</li>
 </ol>
 </section>
 </section>

notebooks/Sample-Hardness.html

@@ -751,7 +751,7 @@ <h2 class="anchored" data-anchor-id="margins">Margins</h2>
 </code></pre>
 </div>
 </div>
-<p>Overall accuracy improves. Based on our earlier observations, we can predict class 1 much better than class 0.</p>
+<p>Overall accuracy improves. Based on our earlier observations, we can predict class 1 much better than class 0. Interestingly, this can also be interpreted as a different form of regularization. Typically, one would place a constraint on the norm of the parameters, implying, one is enforcing smoothness constraints on the functional space. Here, by reweighting the loss, the learning algorithm gives less importance is difficulty samples, there by, the function to be fit, need to do lot of hard work (i.e very complex function) but a simpler function (meaning smooth function) would suffice. So, while the goal is same (smooth function), the way one goes about can be different. The path of regularization, to a large extent, is a brute-force approach, but reweighting one exactly knowns what is the influence of each example in the training.</p>
 
 
 </section>