Z SingerinTowards Data ScienceBeyond the Derivative — SubderivativesMachine Learning relies heavily on differentiable functions, but that’s rarely the case in practice. What can we do instead?Aug 29, 20191Aug 29, 20191
Z SingerinTowards Data ScienceTopological Data Analysis — Unpacking the BuzzwordSeriously, what the heck is Topological Data Analysis? A concrete explanation with pictures and words.Jun 25, 20192Jun 25, 20192
Z SingerinTowards Data ScienceBasics of Independent Component AnalysisA decomposition tool that humans are naturally good at, but don’t think about.Jun 22, 20192Jun 22, 20192
Z SingerinTowards Data Science3 Quick Tips for Training ML ModelsYour model won’t converge. Help.Jun 18, 2019Jun 18, 2019
Z SingerinTowards Data ScienceSoftmax and UncertaintyThe softmax function carries a probabilistic interpretation in neural networks. When does this interpretation break down and mislead…May 6, 20191May 6, 20191
Z SingerinTowards Data ScienceWhy Norms Matters — Machine LearningThere are countless names and uses for the L¹ and L² norms — but what do they really represent and how are they different?Mar 25, 20193Mar 25, 20193
Z SingerinTowards Data ScienceUnderstanding Machine Learning on Point Clouds through PointNet++Point clouds are a convenient way of representing spatial data and other unordered data. But what are they, and how are they used in ML?Jan 25, 20194Jan 25, 20194
Z SingerinTowards Data ScienceManifolds in Data Science — A Brief OverviewManifolds are an essential tool for representing data in higher dimensions. But what are manifolds, and how are they used?Jan 7, 201910Jan 7, 201910
Z SingerinTowards Data ScienceUnderstanding High Dimensional Spaces in Machine LearningHigh dimensional spaces are everywhere in machine learning. How can we think about these spaces in a concrete way?Dec 28, 20182Dec 28, 20182