Practical AI - Building a Recommendation System
1 like341 views
Practical AI/Machine Learning discussion by building a Recommendation System using Collaborative Filtering via Matrix Factorization.
Practical AI - Building a Recommendation System
- 1. Practical AI - Building a Recommendation System John Paul Ada
- 3. Machines built for simulating human intelligence.
- 5. Learning without being told how to.
- 6. ML Problems Here are some types of problems Machine Learning is trying to solve. Regression Classification Clustering Recommendation
- 7. ML Categories Here are some of the current Machine Learning categories. Supervised Learning Unsupervised Learning Semi-Supervised Learning Reinforcement Learning
- 10. Recommending something based on its features.
- 12. Recommending based on past customer ratings.
- 14. Finding smaller matrices whose product would result in the completed target matrix.
- 15. Latent Factors
- 16. The features of the object that you dont know about.
- 17. Matrix Factorization 1. Create two matrices with random values. Each matrix should have the same number of rows as there are users and as many columns as your preferred number of latent factors. 2. Adjust the two matrices until their product nearly matches the target matrix (gradient descent). 3. Multiply the matrices to get the completed matrix.
- 18. Factor Matrices
- 19. Completed Matrix
- 20. Questions?
- 21. Exercise