Welcome to my sandbox for all things classification! This repo is where I tinker with machine learning models that assign objects to categories — whether it's pixels in an image, regions in a scan, or features in a dataset.
Classification is basically the art of decision-making based on features. You feed a model some labeled data, it learns the boundaries between classes, and then it tries to guess the label of new stuff. Simple in theory — but in practice, choosing the right features can be a real time sink.
That’s why I built this repo: to experiment with ways to simplify feature selection and automate the process of picking the best feature set for building accurate classifiers. Think of it as my personal R&D lab for smarter, faster ML workflows.
Here’s a mix of classic and modern classifiers I’ve explored:
- 🔍 Support Vector Machines (SVM)
- 🧭 k-Nearest Neighbors (kNN)
- 🧠 Multi-Layer Perceptrons (MLP)
- 🎲 Gaussian Mixture Models (GMM)
- 🕸️ Convolutional Neural Networks (CNNs)
One of the big goals here is to reduce the manual grind of feature engineering. I’ve built and tested tools that help automatically select the most relevant features — so you can spend less time guessing and more time building.
Classification powers real-world applications like:
- 🖼️ Image Segmentation: breaking images into regions of similar color or texture
- 🔍 Object Recognition: identifying specific objects within a mix
- ✅ Quality Control: deciding if something’s good or defective
- 🆕 Novelty Detection: spotting changes or anomalies
- 🔡 Optical Character Recognition (OCR): turning images of text into actual text
Classic dataset, but I’ve added my own twist with feature selection experiments and model comparisons.
From raw pixels to predictions — testing CNNs and MLPs with automated feature selection.
Inspired by real-world manufacturing — precision, recall, and anomaly detection are key.
This repo is a work in progress — and I’m always open to feedback, ideas, or collaboration. If you’re into ML, feature selection, or just curious about classification, feel free to reach out or fork away!
Thanks for checking it out! 🚀