Visão Geral

O Curso Recommender Systems, foi elaborado como uma introdução abrangente aos Sistemas de Recomendação. É necessário um bom entendimento do aprendizado de máquina básico e um entendimento razoável da álgebra linear.

Pre-Requisitos

Aprendizado profundo básico: neurônios, tipos de camadas, redes, funções de perda, otimizadores, overfitting, Tensorflow

Materiais

Conteúdo Programatico

Recommender Systems and where to find them

1. Google
2. Ads
3. Netflix

Basic Recommender Systems

1. Cosine Distance
2. SVD
3. SVD for recommendation system factorization

Hands-on Lab:

• Implement Singular Value Decomposition
• Understand how it affects the Cosine Distance

Candidate Generation

1. Content Based Filtering
2. Collaborative Filtering
3. Matrix Factorization

Hands-on Lab: Use the studied matrix factorization techniques to implement a basic recommendation system and benchmark the factorization methods
Recommender using Deep Neural Networks

1. Softmax Model
2. Softmax Embedding
3. Embeddings for Neural Networks
4. Item2Vec

Hands-on Lab: Build a recommendation system using neural networks embeddings for IMDB movies
Ranking

1. Retrieval
2. Scoring
3. Re-ranking

Hands-on Lab:

• Learn how to improve your previous model by understanding how the movie ranking is made
• Implement your Collaborative Filtering and Hybrid Collaborative Filtering methods to make a better recommendation system

Autoencoder for Recommender Systems

1. Autoencoders
2. Latent Spaces
3. Variational Autoencoders

Hands-on Lab: Build a recommendation system for retail stores using Autoencoders