Arrrrv·Sep 30Benchmarking Classification Models on the UNSW-NB15 Network Intrusion DatasetIntroduction
Lorenzo Vitali·Sep 21Frauds and why you do not want to use all your labelsFor a long time I treated a fraud label the same way I treated any other column: it’s either `0` or `1`, end of story. What I was missing…
Stat@Sovan·Jul 31Which one should we optimize? Statistically most common or practically most important!Most binary classifiers minimize the expected error (risk). Traditional classifiers solve
Lorenzo Vitali·Jul 9Inspiration for some Initial Feature Elimination in classical MLI remember playing with the Titanic (or any other similar) dataset early in my career, doing a few plots, and jumping straight to model…
Lorenzo Vitali·Jun 20Classic Binary Classification Metrics and top-k constraintsWhen I started with binary classifiers I thought the model magically would spit out only a 0 or 1. From those outputs a few simple metrics…
Kalash Sharma·May 23Build & Deploy a Spam Email Detection SystemLet’s start with an important question: Why do we need machine learning to detect spam emails?
Rohit Mugalya·May 1AI that plays Chrome Dino when offlineThe source code and implementation are available in my GitHub repository. Visit RohitMugalya/Chrome-Dino-AI to see the code.
Shahrudra·Apr 27Predicting Forest Fires with Machine Learning: A Deep Dive into the Algerian Forest Fires DatasetForest fires are devastating events that cause immense ecological and economic damage. But what if we could reliably predict them using…
Gajera Riddhi·Mar 3From Using To I Built My First AI Model: To Predict Incidents To Prevent Outages Before They Happen
Infinite Ripples·Feb 6A Layer-Wise Derivation of Backpropagation in Neural NetworksNeural networks learn by making mistakes—and then systematically correcting them. Every accurate prediction is built upon a chain of…
