This repository contains homework and mini-projects for AI4032 — Introduction to Artificial Intelligence / Machine Learning.
It also links to the Final Project hosted separately.

HW1/

Mini projects are organized in dedicated folders.

Contents

• Ai4032_MP1_40216723.ipynb — full implementation (Colab-ready)
• mp1_lr_dataset_ai4032.csv — linear regression dataset
• Titanic-Dataset.csv — auxiliary dataset
• Titanic.pdf — analysis/report for the Titanic task

Highlights

• Data preprocessing & visualization
• Linear / multivariate regression
• Robust regression (Huber, RANSAC)
• 3D visualization & model evaluation

Contents

• AI4032_MP2_Q1.ipynb — image binarization & pixel ops
• AI4032_MP2_Q2.ipynb — perceptron-based shape detection
• AI4032_MP2_Q3.ipynb — Hamming network with noise
• AI4032_MP2_Q4.ipynb — CNN-based recognition (optional)
• REPORT_AI4032_MP2.pdf — full report & results

Highlights

• Vectorization for Persian characters
• Synthetic noise & missing-point handling
• Hamming network implementation
• Shallow vs. deep method comparison

Contents

• REPORT MP3.pdf — full write-up (link: ./MP3/REPORT%20MP3.pdf)
• Notebooks / source code for all MP3 tasks

What’s inside (overview)

• Fuzzy vs. MLP function approximation (1-D, 2-D)
• Defuzzifier comparison (Centroid / Bisector / MOM)
• Lookup-table prediction for Mackey–Glass time-series
• Sugeno-type fuzzy modeling for a dynamic system
• Simple robot navigation with fuzzy rules
• ANFIS-style clustering & rule extraction (classification)

Folder: Final_Exam_online/
Report: Final_Exam_online.pdf

Includes discrete simulation steps, membership functions, Mamdani inference pipeline (AND=min, weighted average defuzzification), and analysis figures.

Repo: Final Project (Emotion recognition DEAP LSTM SHAP)

End-to-end EEG emotion recognition on DEAP with interpretable modeling (xLSTM/attention + SHAP), dataset handling, training/evaluation scripts, and visualization utilities.

• Colab: open any *.ipynb in Google Colab and run all cells.
• Local: Python ≥3.10 recommended. Create a virtual env, install notebook-specific requirements listed at the top of each notebook, then run.

python -m venv .venv
source .venv/bin/activate  # (Windows: .venv\Scripts\activate)
pip install -r requirements.txt  # if present; otherwise see notebook headers