FFM Tutorial: Field-aware Factorization Machine

FFM Tutorial: Field-aware Factorization Machine#

Sandbox Model — Experimental

This model is part of the CoreRec Sandbox — an experimental collection of algorithms for research and exploration. Sandbox models are not production-tested and may have incomplete implementations. For production-ready models, see the Production Models.

Introduction#

FFM extends FM with field-aware latent vectors where each feature has different embeddings for interacting with different fields.

Paper: Juan et al. 2016 - Field-aware Factorization Machines for CTR Prediction

How FFM Works#

Architecture#

Field-Aware Interactions:

  1. Feature Fields: Group features (e.g., User, Item, Context)

  2. Field-Specific Embeddings: v_{i,f_j} for feature i w.r.t. field j

  3. Pairwise Interactions: Use appropriate field-specific vectors

Key Innovation: Different latent factors per field pair

Example:

  • Publisher × Advertiser uses v_{pub,adv} and v_{adv,pub}

  • Publisher × Gender uses v_{pub,gender} and v_{gender,pub}

Mathematical Foundation#

FFM Formula:

ŷ = w_0 + Σ_i w_i·x_i + Σ_i Σ_{j>i} <v_{i,f_j}, v_{j,f_i}>·x_i·x_j

where:

  • v_{i,f_j}: latent vector for feature i w.r.t. field of feature j

  • f_j: field of feature j

vs Standard FM:

FM:  <v_i, v_j>  # same vector always
FFM: <v_{i,f_j}, v_{j,f_i}>  # field-aware

Reference Implementation#

Implementation Note

The code examples below are reference implementations for learning purposes. This is a sandbox model and has not been production-tested. Verify behavior thoroughly before using in production.

Step 1: Import and Load Data#

from corerec.sandbox.collaborative_full.nn_base.FFM_base import FFM_base as FFM
from cr_learn import ml_1m
from sklearn.model_selection import train_test_split
import numpy as np

# Load dataset with CORRECT API
data = ml_1m.load()  # Returns dict with 'ratings', 'users', 'movies'
ratings_df = data['ratings']  # DataFrame with user_id, movie_id, rating, timestamp

print(f"Loaded {len(ratings_df)} ratings")

# Split data
train_df, test_df = train_test_split(ratings_df, test_size=0.2, random_state=42)

# Extract arrays for model
train_users = train_df['user_id'].values
train_items = train_df['movie_id'].values
train_ratings = train_df['rating'].values

test_users = test_df['user_id'].values
test_items = test_df['movie_id'].values
test_ratings = test_df['rating'].values

Step 2: Initialize Model#

model = FFM(
    name="FFM_Model",
    n_factors=4,
    learning_rate=0.2,
    reg=0.00002,
    epochs=20,
    batch_size=256,
    verbose=True
)

print(f"Initialized {model.name}")

Step 3: Train#

model.fit(
    user_ids=train_users,
    item_ids=train_items,
    ratings=train_ratings
)

print("Training complete!")

Step 4: Predict#

# Single prediction
score = model.predict(user_id=1, item_id=100)
print(f"Predicted score: {score:.3f}")

# Batch predictions
test_predictions = model.batch_predict(list(zip(test_users[:100], test_items[:100])))

Step 5: Recommend#

# Get top-10 recommendations for user
user_id = 1
recommendations = model.recommend(
    user_id=user_id,
    top_k=10
)

print(f"Top-10 recommendations for User {user_id}:")
for rank, item_id in enumerate(recommendations, 1):
    print(f"  {rank}. Item {item_id}")

Step 6: Evaluate#

from sklearn.metrics import mean_squared_error
import numpy as np

# Predict all test ratings
test_pred = [model.predict(u, i) for u, i in zip(test_users, test_items)]
rmse = np.sqrt(mean_squared_error(test_ratings, test_pred))
print(f"Test RMSE: {rmse:.4f}")

Step 7: Save & Load#

# Save model
model.save('artifacts/ffm_model')

# Load model
loaded = FFM.load('artifacts/ffm_model')
test_score = loaded.predict(1, 100)
print(f"Loaded model prediction: {test_score:.3f}")

Key Takeaways#

When to Use FFM#

✅ Perfect For:

  • CTR prediction with field structure

  • Sparse categorical features

  • When feature fields matter (User/Item/Context)

  • Kaggle competitions

  • Display advertising

❌ Not For:

  • No natural field structure

  • Dense numerical features

  • Need deep interactions

  • Very large scale (>10M features)

Best Practices#

  1. Latent Factors: k=4-8 (smaller than FM!)

  2. Optimizer: AdaGrad or FTRL

  3. Learning Rate: 0.1-0.2 (higher than FM)

  4. Regularization: λ = 0.00002 typical

  5. Early Stopping: Monitor validation AUC

  6. Normalization: Normalize continuous features

  7. Field Definition: Carefully design fields

  8. Memory: k×fields larger than FM

Scaling & Production Considerations#

If you plan to move this sandbox model toward production use, consider:

  • Data Volume: Test with realistic dataset sizes; sandbox implementations may not be optimized for large-scale data.

  • Distributed Training: For datasets exceeding single-machine memory, consider wrapping the model in distributed training frameworks (e.g., PyTorch DDP, Horovod).

  • Hyperparameter Tuning: Use systematic search (Optuna, Ray Tune) rather than manual tuning. The defaults in sandbox models are not tuned for any specific domain.

  • Serving Latency: Profile inference time. Some architectures (e.g., attention-based) may need quantization or ONNX export for low-latency serving.

  • Monitoring: Implement A/B testing and online metrics (CTR, engagement, diversity) before replacing an existing system.

  • Validation: Write comprehensive unit tests and integration tests before deploying. Sandbox models do not have CI coverage.

Further Reading#

  • Paper: Juan et al. 2016 - Field-aware Factorization Machines for CTR Prediction

  • GitHub Repository

Contents