BiVAE Tutorial: Bilateral Variational Autoencoder

BiVAE Tutorial: Bilateral Variational Autoencoder#

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#

BiVAE uses variational inference to jointly learn latent representations for both users and items with uncertainty quantification.

Paper: Collaborative Filtering via Variational Autoencoders

How BiVAE Works#

Architecture#

Dual Variational Autoencoder:

  1. User Encoder: q(z_u | r_u) → μ_u, σ_u

  2. Item Encoder: q(z_i | r_i) → μ_i, σ_i

  3. Sampling: z_u ~ N(μ_u, σ_u²), z_i ~ N(μ_i, σ_i²)

  4. Decoder: p(r | z_u, z_i) reconstructs rating

  5. KL Regularization: Towards prior N(0,I)

Key Innovation: Probabilistic latent factors with uncertainty

Architecture:

User → Encode → μ_u, σ_u → Sample → z_u                                           Decode → Rating
Item → Encode → μ_i, σ_i → Sample → z_i /

Mathematical Foundation#

ELBO (Evidence Lower Bound):

L = E_{z~q}[log p(r|z_u,z_i)] - KL(q(z_u,z_i) || p(z_u)p(z_i))

Reconstruction:

log p(r|z_u,z_i) = -||r - f(z_u,z_i)||²

KL Divergence:

KL = -0.5 Σ(1 + log σ² - μ² - σ²)

Reparameterization Trick:

z = μ + σ ⊙ ε,  where ε ~ N(0,I)

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.variational_encoder_base.bivae_base import BiVAE_base as BiVAE
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 = BiVAE(
    name="BiVAE_Model",
    latent_dim=20,
    encoder_dims=[200, 100],
    kl_anneal_rate=0.01,
    epochs=20,
    batch_size=256,
    learning_rate=0.001,
    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/bivae_model')

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

Key Takeaways#

When to Use BiVAE#

✅ Good For:

  • Uncertainty quantification

  • Generative modeling (sample ratings)

  • Cold-start with priors

  • Small-medium datasets

  • Research and experimentation

❌ Not For:

  • Large-scale production (slow)

  • Need interpretability

  • Real-time inference

  • When point estimates suffice

Best Practices#

  1. Latent Dimension: 20-50

  2. KL Annealing: Start with β=0.01 → 1.0

  3. Encoder: [200, 100] typical

  4. Decoder: Symmetric [100, 200]

  5. Activation: Tanh or ReLU

  6. Learning Rate: 0.001 with decay

  7. Warmup: 10-20 epochs before full KL

  8. Batch Size: 128-512

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#

Contents