Published:
An episode from Surely You’re Joking, Mr. Feynman! has stuck with me for years, and I wanted to analyze the math behind it. In the story, Feynman describes an interaction with a Japanese man at a restaurant in Brazil.
Published:
In the real world, often, we would not be solving some problem for the first time. There would already be some existing solution in place and we would like to improve upon the existing solution. The improvement might come in different flavors; it can be expansion of the solution or simply improvement of the solution on some performance metric.
Published:
The following list contains my blogs that are published in Towards Data Science. Mostly, the blogs revolve around machine learning, data science, algorithms, and mathematics.
Published:
The following list contains my blogs that are published in HackerNoon. I was awarded Hacker Noon 2020 Contributor of the Year in algorithms category.
Published:
AUROC is one of the most commonly used metric in binary classification tasks. It stands for Area Under Curve - Receiver Operating Characteristics. It is simply the area under the ROC curve. One way of interpreting AUROC is that it is just the area under the ROC curve obtained by interpolating TPR and FPR at different thresholds. There is also another probabilistic interpretation of AUROC. We will state that and prove why that is the case mathematically.
Published:
The OOD baseline paper titled “A Baseline for Detecting Misclassified and Out-of-Distribution Examples in Neural Networks” is one of the most important research paper in OOD detection literature. Though the paper is not mathematically and technically heavy, it certainly has some concepts that might not be straightforward for beginners. The discussion and/or tutorial of the paper has been lacking in the internet. Hence, to help the beginners dive into the field of out-of-distribution detection and make it easier for beginners to fully understand the concepts of the paper, I have written an explainer blog.