🎨 IFT6390 — Drawing Classification | Kaggle 2nd Place 🥈

Course assignments and Kaggle competition notebooks for IFT3395/6390 — Fundamentals of Machine Learning, Université de Montréal.

---

🏆 Kaggle Competition — 2nd Place

Competition: Classification de dessins — IFT3395/6390 2020

Rank	Team	Score	Entries
🥇 1st	EP	0.91285	10
🥈 2nd	Sami Bahig	0.90638	19
🥉 3rd	Carelessly No Flash	0.90400	14

Private leaderboard calculated on ~70% of test data.

---

🎯 Problem

Multi-class classification of hand-drawn sketches — predicting the category of a drawing from raw image features.

Models explored: CNN, DenseNet, Ensemble methods, SVM, Random Forest, Logistic Regression, Naive Bayes (Bernoulli), AdaBoost, Bagging, Voting Classifiers.

---

🗂️ Notebooks

Notebook	Description
Classifications_de_dessins.ipynb	Main classification pipeline
CNN_DenseNet.ipynb	CNN + DenseNet architecture
Ensemble.ipynb	Ensemble methods
VotingFinal.ipynb	Final voting classifier
Multi_SVM_.ipynb	Multi-class SVM
LogisticRegression...ipynb	LR + RF + Bagging + AdaBoost
MultiBernoulliAhmed.ipynb	Naive Bayes Bernoulli
Hw1PratiqueIFT6390.ipynb	Assignment 1
Assignment_2_4_1,_4_2.ipynb	Assignment 2

---

🛠️ Tech Stack

Component	Technology
Language	Python
ML Framework	scikit-learn
Deep Learning	PyTorch / Keras
Environment	Google Colab
Competition	Kaggle

---

👤 Author

Sami Bahig, MD MSc — Data Scientist & AI Engineer IFT3395/6390 — Université de Montréal · 2020

---

MIT License · Sami Bahig · 2020