Bleeding edge alert: Sparse linear methods (SLIM) - Python Tutorial

Contents

• 1. Getting Started 1. Getting Started

  • Install Anaconda, review course materials, and create movie recommendations

    9m 5s
  • Course roadmap

    3m 14s
  • Understanding you through implicit and explicit ratings

    4m 25s
  • Top-N recommender architecture

    5m 53s
  • Review the basics of recommender systems

    4m 46s
• 2. Introduction to Python 2. Introduction to Python

  • Data structures in Python

    5m 17s
  • Functions in Python

    2m 46s
  • Booleans, loops, and a hands-on challenge

    3m 52s
• 3. Evaluating Recommender Systems 3. Evaluating Recommender Systems

  • Train/test and cross-validation

    3m 49s
  • Accuracy metrics (RMSE and MAE)

    4m 6s
  • Top-N hit rate: Many ways

    4m 35s
  • Coverage, diversity, and novelty

    4m 55s
  • Churn, responsiveness, and A/B tests

    5m 6s
  • Review ways to measure your recommender

    2m 55s
  • Walkthrough of RecommenderMetrics.py

    6m 53s
  • Walkthrough of TestMetrics.py

    5m 8s
  • Measure the performance of SVD recommendations

    2m 24s
• 4. A Recommender Engine Framework 4. A Recommender Engine Framework

  • Our recommender engine architecture

    7m 27s
  • Recommender engine walkthrough, part 1

    3m 55s
  • Recommender engine walkthrough, part 2

    3m 51s
  • Review the results of our algorithm evaluation

    3m 11s
• 5. Content-Based Filtering 5. Content-Based Filtering

  • Content-based recommendations and the cosine similarity metric

    8m 58s
  • K-nearest neighbors (KNN) and content recs

    3m 59s
  • Producing and evaluating content-based movie recommendations

    5m 23s
  • Bleeding edge alert: Mise-en-scene recommendations

    4m 31s
  • Dive deeper into content-based recommendations

    4m 26s
• 6. Neighborhood-Based Collaborative Filtering 6. Neighborhood-Based Collaborative Filtering

  • Measuring similarity and sparsity

    4m 49s
  • Similarity metrics

    8m 32s
  • User-based collaborative filtering

    7m 25s
  • User-based collaborative filtering: Hands-on

    4m 59s
  • Item-based collaborative filtering

    4m 14s
  • Item-based collaborative filtering: Hands-on

    2m 23s
  • Tuning collaborative filtering algorithms

    3m 31s
  • Evaluating collaborative filtering systems offline

    1m 28s
  • Measure the hit rate of item-based collaborative filtering

    2m 17s
  • KNN recommenders

    4m 4s
  • Running user- and item-based KNN on MovieLens

    2m 26s
  • Experiment with different KNN parameters

    4m 25s
  • Bleeding edge alert: Translation-based recommendations

    2m 29s
• 7. Matrix Factorization Methods 7. Matrix Factorization Methods

  • Principal component analysis (PCA)

    6m 31s
  • Singular value decomposition (SVD)

    7m 6s
  • Running SVD and SVD++ on MovieLens

    3m 46s
  • Improving on SVD

    4m 33s
  • Tune the hyperparameters on SVD

    1m 58s
  • Bleeding edge alert: Sparse linear methods (SLIM)

    3m 30s
• 8. Introduction to Deep Learning 8. Introduction to Deep Learning

  • Deep learning introduction

    1m 30s
  • Deep learning prerequisites

    8m 13s
  • History of artificial neural networks

    10m 51s
  • Playing with TensorFlow

    12m 2s
  • Training neural networks

    5m 47s
  • Tuning neural networks

    3m 52s
  • Introduction to TensorFlow

    11m 29s
  • Handwriting recognition with TensorFlow, part 1

    13m 18s
  • Handwriting recognition with TensorFlow, part 2

    12m 3s
  • Introduction to Keras

    2m 48s
  • Handwriting recognition with Keras

    9m 52s
  • Classifier patterns with Keras

    3m 58s
  • Predict political parties of politicians with Keras

    9m 55s
  • Intro to convolutional neural networks (CNNs)

    8m 59s
  • CNN architectures

    2m 54s
  • Handwriting recognition with CNNs

    8m 38s
  • Intro to recurrent neural networks (RNNs)

    7m 38s
  • Training recurrent neural networks

    3m 21s
  • Sentiment analysis of movie reviews using RNNs and Keras

    11m 1s
• 9. Deep Learning for Recommender Systems 9. Deep Learning for Recommender Systems

  • Intro to deep learning for recommenders

    2m 19s
  • Restricted Boltzmann machines (RBMs)

    8m 2s
  • Recommendations with RBMs, part 1

    12m 46s
  • Recommendations with RBMs, part 2

    7m 11s
  • Evaluating the RBM recommender

    3m 44s
  • Tuning restricted Boltzmann machines

    1m 43s
  • Exercise results: Tuning a RBM recommender

    1m 15s
  • Auto-encoders for recommendations: Deep learning for recs

    4m 27s
  • Recommendations with deep neural networks

    7m 23s
  • Clickstream recommendations with RNNs

    7m 23s
  • Get GRU4Rec working on your desktop

    2m 42s
  • Exercise results: GRU4Rec in action

    7m 51s
  • Bleeding edge alert: Deep factorization machines

    5m 49s
  • More emerging tech to watch

    5m 14s
• 10. Scaling It Up 10. Scaling It Up

  • Introduction and installation of Apache Spark

    5m 49s
  • Apache Spark architecture

    5m 13s
  • Movie recommendations with Spark, matrix factorization, and ALS

    6m 2s
  • Recommendations from 20 million ratings with Spark

    4m 57s
  • Amazon DSSTNE

    4m 41s
  • DSSTNE in action

    9m 25s
  • Scaling up DSSTNE

    2m 14s
  • AWS SageMaker and factorization machines

    4m 24s
  • SageMaker in action: Factorization machines on one million ratings, in the cloud

    7m 39s
• 11. Real-World Challenges of Recommender Systems 11. Real-World Challenges of Recommender Systems

  • The cold start problem (and solutions)

    6m 12s
  • Implement random exploration

    54s
  • Exercise solution: Random exploration

    2m 18s
  • Stoplists

    4m 48s
  • Implement a stoplist

    32s
  • Exercise solution: Implement a stoplist

    2m 22s
  • Filter bubbles, trust, and outliers

    5m 39s
  • Identify and eliminate outlier users

    44s
  • Exercise solution: Outlier removal

    4m
  • Fraud, the perils of clickstream, and international concerns

    4m 33s
  • Temporal effects and value-aware recommendations

    3m 30s
• 12. Case Studies 12. Case Studies

  • Case study: YouTube, part 1

    3m 42s
  • Case study: YouTube, part 2

    7m 4s
  • Case study: Netflix, part 1

    3m 59s
  • Case study: Netflix, part 2

    3m 55s
• 13. Hybrid Approaches 13. Hybrid Approaches

  • Hybrid recommenders and exercise

    2m 54s
  • Exercise solution: Hybrid recommenders

    4m 17s
• Conclusion Conclusion

  • More to explore

    2m 31s