Recommendation system using deep learning (Day 2)

Description

In the digital world, recommendation systems play a significant role—both for the users and for the company. For users, it opens up a new world of options that were previously tough to find. For companies, it helps drive user engagement and satisfaction, directly impacting the bottom line. If you’ve shopped on an ecommerce site or watched a movie on an on-demand video platform, you’ve seen options like, “People who viewed this product also viewed…” or, “Products similar to this one.…” These are the results from recommendation systems (recsys).

Amit Kapoor and Bargava Subramanian walk you through the different paradigms of recommendation systems and introduced you to deep learning-based approaches. You’ll leave with the practical hands-on knowledge to build, select, deploy, and maintain a recommendation system.

Outline

Introduction

• Why build recommendation systems? Scope and evolution of recsys
• Challenges: Accuracy, ranking, relevance, novelty, serendipity, and diversity
• Paradigms in recommendations: Content based, collaborative filtering, knowledge based, hybrid, and ensemble
• Key concepts in recsys: Explicit versus implicit feedback, user-item matrix, domain signals (location, time, context, and social)
• Why use deep learning for recsys? Traditional versus deep learning approaches
• Primer on deep learning with examples and use cases

Content based

• Introduction to the case: Product recommendation
• Environment setup for hands-on session
• Feature extraction using deep learning: Embeddings for heterogeneous data
• Exercise: Recommend items using similarity measures

Collaborative filtering

• Overview of traditional collaborative filtering for recsys
• Primer on deep learning approaches: Deep matrix factorization and autoencoders
• Exercise: Recommend items using collaborative filtering

Learning to rank

• Why learning to rank? Prediction versus ranking
• Rank learning approaches: Pointwise, pairwise, and listwise
• Deep learning approach to combine prediction and ranking
• Exercise: Recommend items using learning to rank

Hybrid recommender

• Introduction to the case: Text recommendation
• Combining content-based and collaborative filtering
• Primer on wide and deep learning for recommender systems
• Exercise: Recommend items using hybrid recommender

Time and context

• Adding temporal component: Window and decay based
• Adding context through group recommendations
• Dynamic and sequential modeling using recurrent neural networks (RNNs)
• Exercise: Recommend items using RNN recommender

Deployment and monitoring

• Deploying the recommendation system models
• Measuring improvements from recommendation system
• Improving the models based on the feedback from production
• Architecture design for recsys: Offline, nearline, and online

Evaluation, challenges, and the way forward

• A/B testing for recommendation systems
• Challenges in recsys: Building explanations, model debugging, scaling out and up, fairness, accountability, and trust
• Bias in recsys: Training data, UI → Algorithm → UI, private
• When not to use deep learning for recsys
• Recap, next steps, and learning resources