🧠 Branch Summary

Branch	Description
main	Baseline Fuzzy-Neural CF with pipeline and hyperparameter tuning
gaussian_fuzzy_cf_v2	Improved Gaussian membership Fuzzy CF with Huber loss and deeper architecture

🎵 Fuzzy Neural Collaborative Filtering (MovieLens 100K)

This repository implements the Fuzzy Neural Collaborative Filtering (Fuzzy-Neural CF) model inspired by the 2024 study, combining fuzzy logic and neural collaborative filtering to enhance recommendation accuracy.

---

🧠 Abstract

This work reimplements the Fuzzy-Neural CF method using the MovieLens 100K dataset.
It fuses four fuzzy membership degrees — positive (μ), neutral (η), negative (ν), and refusal (ω) — into a neural network that learns nonlinear relationships for user–item preference prediction.

---

🧱 Project Structure

fuzzy_neural_cf/ ├── data/ ├── src/ │ ├── preprocessing/ │ ├── models/ │ └── utils/ ├── results/ └── notebooks/

---

⚙️ Setup & Installation

# Clone the repo
git clone https://github.com/<your-username>/fuzzy_neural_cf.git
cd fuzzy_neural_cf

# Install dependencies using uv
uv sync

🚀 Training Pipeline

python src/preprocessing/build_preference_matrices.py
python src/models/train_bemf.py
python src/models/compute_memberships.py
python src/models/fuzzy_cf.py