Building Recommender Systems with Machine Learning and AI

Building Recommender Systems with Machine Learning and AI

by Frank Kane
Released January 2021
Publisher(s): Packt Publishing
ISBN: 9781789803273

Read it now on the O’Reilly learning platform with a 10-day free trial.

O’Reilly members get unlimited access to books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.

Start your free trial

Video description

This course will teach you how to use Python, artificial intelligence (AI), machine learning, and deep learning to build a recommender system. From creating a simple recommendation engine to building hybrid ensemble recommenders, you will learn key concepts effectively and in a real-world context.

The course starts with an introduction to the recommender system and Python. Learn how to evaluate recommender systems and explore the architecture of the recommender engine framework. Next, you will learn to understand how content-based recommendations work and get to grips with neighborhood-based collaborative filtering. Moving along, you will learn to grasp model-based methods used in recommendations, such as matrix factorization and Singular Value Decomposition (SVD).

Next, you will learn to apply deep learning, artificial intelligence (AI), and artificial neural networks to recommendations and learn how to scale massive datasets with Apache Spark machine learning. Later, you will encounter real-world challenges of recommender systems and learn how to solve them. Finally, you will study the recommendation system of YouTube and Netflix and find out what a hybrid recommender is.

By the end of this course, you will be able to build real-world recommendation systems that will help users discover new products and content online.

What You Will Learn

  • Get a basic overview of the architecture of recommender systems
  • Test and evaluate recommendation algorithms with Python
  • Use K-Nearest-Neighbors to recommend items to users
  • Find solutions to common issues with large-scale recommender systems
  • Make session-based recommendations with recurrent neural networks
  • Use Apache Spark to compute recommendations at a large scale on a cluster

Audience

This course is suitable for software developers, engineers, and computer scientists who are looking to build recommender systems using the principles of machine learning, deep learning, and artificial intelligence (AI). A basic understanding of Python programming and algorithms is needed to get started with this course.

About The Author

Frank Kane: Frank Kane has spent nine years at Amazon and IMDb, developing and managing the technology that automatically delivers product and movie recommendations to hundreds of millions of customers all the time. He holds 17 issued patents in the fields of distributed computing, data mining, and machine learning. In 2012, Frank left to start his own successful company, Sundog Software, which focuses on virtual reality environment technology and teaches others about big data analysis.

Publisher resources

Download Example Code

Table of contents

  1. Chapter 1 : Getting Started
    1. Installing Anaconda, Course Materials, and Creating Movie Recommendations
    2. Course Roadmap
    3. What Is a Recommender System?
    4. Types of Recommenders
    5. Understanding the Implicit and Explicit Ratings
    6. Top-N Recommender Architecture
    7. Review the Basics of a Recommender System
  2. Chapter 2 : Introduction to Python
    1. The Basics of Python (Activity)
    2. Data Structures in Python
    3. Functions in Python
    4. Booleans, Loops, and a Hands-on Challenge (Exercise)
  3. Chapter 3 : Evaluating a Recommender System
    1. Train/Test and Cross-Validation
    2. Accuracy Metrics Using Root Mean Squared Error (RMSE) and Mean Absolute Error (MAE)
    3. Top-N Hit Rate - Many Ways
    4. Coverage, Diversity, and Novelty
    5. Churn, Responsiveness, and A/B Tests
    6. Reviewing Ways to Measure Your Recommender (Quiz)
    7. Walkthrough of RecommenderMetrics.py (Activity)
    8. Walkthrough of TestMetrics.py (Activity)
    9. Measuring the Performance of Singular Value Decomposition (SVD) Recommendations (Activity)
  4. Chapter 4 : A Recommender Engine Framework
    1. Recommender Engine Architecture
    2. Recommender Engine Walkthrough – Part 1 (Activity)
    3. Recommender Engine Walkthrough – Part 2 (Activity)
    4. Reviewing the Results of Our Algorithm Evaluation (Activity)
  5. Chapter 5 : Content-Based Filtering
    1. Content-Based Recommendations and the Cosine Similarity Metric
    2. K-Nearest-Neighbors (KNN) and Content Recommendations
    3. Producing and Evaluating Content-Based Movie Recommendations (Activity)
    4. A Note on Using Implicit Ratings
    5. Bleeding Edge Alert! Mise-En-Scene Recommendations (Activity)
    6. Diving Deep into Content-Based Recommendations (Exercise)
  6. Chapter 6 : Neighborhood-Based Collaborative Filtering
    1. Measuring Similarity and Sparsity
    2. Similarity Metrics
    3. User-Based Collaborative Filtering
    4. User-Based Collaborative Filtering - Hands-On (Activity)
    5. Item-Based Collaborative Filtering
    6. Item-Based Collaborative Filtering - Hands-On (Activity)
    7. Tuning Collaborative Filtering Algorithms (Exercise)
    8. Evaluating Collaborative Filtering Systems Offline (Activity)
    9. Measuring the Hit Rate of Item-based Collaborative Filtering (Exercise)
    10. K-Nearest-Neighbor (KNN) Recommenders
    11. Running User-Based and Item-Based K-Nearest-Neighbor (KNN) on MovieLens Dataset (Activity)
    12. Experimenting with Different K-Nearest-Neighbor (KNN) Parameters (Exercise)
    13. Bleeding Edge Alert! Translation-Based Recommendations
  7. Chapter 7 : Matrix Factorization Methods
    1. Principal Component Analysis (PCA)
    2. Singular Value Decomposition (SVD)
    3. Running Singular Value Decomposition (SVD) and Singular Value Decomposition (SVD ++) on MovieLens (Activity)
    4. Improving on Singular Value Decomposition (SVD)
    5. Tuning the Hyperparameters on Singular Value Decomposition (SVD) (Exercise)
    6. Bleeding-Edge Alert! Sparse Linear Methods (SLIM)
  8. Chapter 8 : Introduction to Deep Learning (Optional)
    1. Deep Learning Introduction
    2. Deep Learning Prerequisites
    3. History of Artificial Neural Networks (ANN)
    4. (Activity) Playing with TensorFlow
    5. Training Neural Networks
    6. Tuning Neural Networks (Avoiding Overfitting)
    7. Activation Functions: More Depth
    8. Introduction to TensorFlow
    9. Handwriting Recognition with TensorFlow – Part 1 (Activity)
    10. Handwriting Recognition with TensorFlow – Part 2 (Activity)
    11. Introduction to Keras
    12. Handwriting Recognition with Keras (Activity)
    13. Classifier Patterns with Keras
    14. Predicting Political Parties of Politicians with Keras (Exercise)
    15. Introduction to Convolutional Neural Network (CNN)
    16. Convolutional Neural Network (CNN) Architecture
    17. Handwriting Recognition with Convolutional Neural Network (CNN) (Activity)
    18. Introduction to Recurrent Neural Networks (RNN)
    19. Training Recurrent Neural Networks (RNN)
    20. Sentiment Analysis of Movie Reviews Using Recurrent Neural Networks (RNN) and Keras (Activity)
    21. Tuning Neural Networks
    22. Neural Network Regularization Techniques
    23. Generative Adversarial Networks (GANs)
    24. GANs in Action
    25. Generating Images of Clothing with Generative Adversarial Networks (Activity)
  9. Chapter 9 : Deep Learning for Recommender Systems
    1. Introduction to Deep Learning for Recommenders
    2. Restricted Boltzmann Machine (RBM)
    3. Recommendations with Restricted Boltzmann Machine (RBM) – Part 1 (Activity)
    4. Recommendations with Restricted Boltzmann Machine (RBM) – Part 2 (Activity)
    5. Evaluating the Restricted Boltzmann Machine (RBM) Recommender (Activity)
    6. Tuning Restricted Boltzmann Machine (RBM) (Exercise)
    7. Exercise Results: Tuning a Restricted Boltzmann Machine (RBM) Recommender
    8. Auto-Encoders for Recommendations: Deep Learning for Recommendation
    9. Recommendations with Deep Neural Networks (Activity)
    10. Clickstream Recommendations with Recurrent Neural Networks (RNN)
    11. Getting GRU4Rec to Work on Your Desktop (Exercise)
    12. Exercise Results: GRU4Rec in Action
    13. Bleeding Edge Alert! Generative Adversarial Networks for Recommendations
    14. TensorFlow Recommenders (TFRS): Introduction and Building a Retrieval Stage
    15. TensorFlow Recommenders (TFRS): Building a Ranking Stage
    16. TensorFlow Recommenders (TFRS): Incorporating Side Features and Deep Retrieval
    17. TensorFlow Recommenders (TFRS): Multi-Task Recommenders, Deep and Cross Networks, ScaNN, and Serving
    18. Bleeding Edge Alert! Deep Factorization Machines
    19. More Emerging Tech to Watch
  10. Chapter 10 : Scaling It Up
    1. Introduction and Installation of Apache Spark (Activity)
    2. Apache Spark Architecture
    3. Movie Recommendations with Spark, Matrix Factorization, and Alternating Least Squares (ALS) (Activity)
    4. Recommendations from 20 Million Ratings with Spark (Activity)
    5. Amazon Deep Scalable Sparse Tensor Network Engine (DSSTNE)
    6. Amazon Deep Scalable Sparse Tensor Network Engine (DSSTNE) in Action
    7. Scaling Up Amazon Deep Scalable Sparse Tensor Network Engine (DSSTNE)
    8. Amazon Web Services (AWS) SageMaker and Factorization Machines
    9. Amazon SageMaker in Action: Factorization Machines on One Million Ratings in the Cloud
    10. Other Systems of Note (Amazon Personalize, RichRelevance, Recombee, and More)
    11. Recommender System Architecture
  11. Chapter 11 : Real-World Challenges of Recommender Systems
    1. The Cold Start Problem (and Solutions)
    2. Implementing Random Exploration (Exercise)
    3. Exercise Solution – Random Exploration
    4. Stoplists
    5. Implementing a Stoplist (Exercise)
    6. Exercise Solution – Implementing a Stoplist
    7. Filtering Bubbles, Trust, and Outliers
    8. Identifying and Eliminating Outlier Users (Exercise)
    9. Exercise Solution: Outlier Removal
    10. Fraud, the Perils of Clickstream, and International Concerns
    11. Temporal Effects and Value-Aware Recommendations
  12. Chapter 12 : Case Studies
    1. Case Study: YouTube – Part 1
    2. Case Study: YouTube – Part 2
    3. Case Study: Netflix – Part 1
    4. Case Study: Netflix – Part 2
  13. Chapter 13 : Hybrid Approaches
    1. Hybrid Recommenders and Exercise
    2. Exercise Solution: Hybrid Recommenders
  14. Chapter 14 : Wrapping Up
    1. More to Explore

Product information

  • Title: Building Recommender Systems with Machine Learning and AI
  • Author(s): Frank Kane
  • Release date: January 2021
  • Publisher(s): Packt Publishing
  • ISBN: 9781789803273