AutoInt Tutorial: Automatic Feature Interaction Learning via Self-Attention

AutoInt Tutorial: Automatic Feature Interaction Learning via Self-Attention#

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#

AutoInt uses multi-head self-attention to automatically learn high-order feature interactions without manual feature engineering.

Paper: Song et al. 2019 - AutoInt: Automatic Feature Interaction Learning via Self-Attentive Neural Networks

How AutoInt Works#

Architecture#

Self-Attention for Feature Interactions:

  1. Embedding Layer: Convert sparse features to dense embeddings

  2. Interacting Layer: Multi-head self-attention on feature embeddings

    • Each head learns different interaction patterns

    • Residual connections preserve original features

  3. Stacking: Multiple attention layers for higher-order interactions

  4. Output Layer: Combine all layers for prediction

Key Innovation: Treats features as a sequence, applies self-attention

Architecture:

Features → Embed → [Self-Attention + Residual]×L → Concat → Output

Mathematical Foundation#

Multi-Head Self-Attention:

head_h = Attention(E·W_h^Q, E·W_h^K, E·W_h^V)
where E = [e_1, e_2, ..., e_m]  # feature embeddings

Attention Mechanism:

Attention(Q,K,V) = softmax((Q·K^T)/√d) · V

Multi-Head Output:

MultiHead(E) = [head_1; head_2; ...; head_H] × W^O

Residual Connection:

E' = ReLU(MultiHead(E) + E)

Final Prediction:

ŷ = σ(w^T · [E^(0); E^(1); ...; E^(L)])

where E^(l) is output of layer l

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.AutoInt_base import AutoInt_base as AutoInt
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 = AutoInt(
    name="AutoInt_Model",
    embedding_dim=32,
    attention_dim=32,
    num_heads=2,
    num_layers=3,
    dropout=0.1,
    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/autoint_model')

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

Key Takeaways#

When to Use AutoInt#

✅ Excellent For:

  • Feature-rich CTR prediction

  • When manual feature engineering is expensive

  • Discovering unknown feature interactions

  • Sparse categorical features

  • Display advertising, app recommendations

❌ Not For:

  • Simple datasets (overkill)

  • No features available (use CF)

  • Sequential patterns (use RNN)

  • Very large feature spaces (memory intensive)

Best Practices#

  1. Attention Heads: 2-4 heads sufficient

  2. Layers: 2-3 interacting layers

  3. Embedding Dim: 16-64 per feature

  4. Attention Dim: Same as embedding

  5. Residual: Always use residual connections

  6. Dropout: 0.1-0.2 on attention weights

  7. Learning Rate: 0.001 with warmup

  8. Batch Size: 256-1024

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: Song et al. 2019 - AutoInt: Automatic Feature Interaction Learning via Self-Attentive Neural Networks

  • GitHub Repository

Contents