Essa é uma revisão anterior do documento!
CE-064: Introdução ao Machine Learning
Detalhes da oferta da disciplina
- Curso: Estatística.
 - Período: 2016/1.
 - Horários e Locais:
- Terça-feira 20:40 - 22:00 (LABEST)
 - Quinta-feira 17:30 - 19:00 (LABEST)
 
 - Atendimento: Sexta, 17h30-18h30.
 
Avaliações
Aulas
| Assunto | Teoria | Prática | Sabatina | Material complementar | 
|---|---|---|---|---|
| Introdução | Aula 1 | Statistical Modeling: The Two Cultures | ||
| Regressão Linear - Revisão | Aula 2 | Prática 2 | Exemplo 2 | When do you use machine learning vs. statistical regression? | 
| Gradiente descendente | Aula 3 | Prática 3 | Exemplo 3 | Stochastic Gradient Descent Tricks | 
| Regularização | Aula 4 | Prática 4 | Exemplo 4 | Regression shrinkage and selection via the lasso | 
| Métodos de reamostragem | Aula 5 | Prática 5 | Exemplo 5 | Cross-validation | 
| Classificação | Aula 6 | Prática 6 | Exemplo 6 | Regularized Discriminant Analysis | 
| Support Vector Machines | Aula 7 | Prática 7 | Exemplo 7 | Learning with kernels | 
| Métodos baseados em árvores | Aula 8 | Prática 8 | An Empirical Comparison of Supervised Learning Algorithms | |
| Manipulação de textos e imagens | Aula 9 | Dissertacao Larissa Sayuri | ||
| Aprendizado não supervisionado | Brief Review of Regression-Based and Machine Learning Methods | 
Notas
Programa da disciplina
- Introdução:
- Aprendizado supervisionado e não supervisionado;
 - Função custo;
 - Revisão de Álgebra matricial.
 
 - Regressão Linear:
- Regressão linear simples e múltipla;
 - Estimação dos parâmetros;
 - Preditor quantitativo e qualitativo;
 - Interpretação.
 
 - Gradiente descendente:
- Batch;
 - Stochastic;
 - Boosting.
 
 - Seleção de modelos e regularização:
- Seleção por Forward e Backward;
 - Estimação do erro do teste utilizando Mallow’s Cp, AIC, BIC, R quadrado ajustado;
 - Regressão Ridge;
 - Regressão Lasso;
 - Horseshoe;
 - Elastic net.
 
 - Métodos de reamostragem:
- Estimando o erro de previsão;
 - Validação cruzada;
 - Bootstrap.
 
 - Classificação:
- Regressão logística;
 - Análise de discriminante linear;
 - Análise de discriminante quadrática;
 - Naive Bayes.
 
 - Support Vector Machines:
- Maximal Margin Classifier;
 - Support vector Classifier;
 - Kernels.
 
 - Métodos baseados em árvores:
- Árvores de decisão;
 - Randon Forest;
 - Bootstrap Aggregation (Bagging);
 - Boosting.
 
 - Aprendizado não supervisionado:
- Análise de componentes principais;
 - K-means Clustering;
 - Hierarchical Clustering.
 
 - Manipulação de textos e imagens.
 
Referências Bibliográficas
* James, G., Witten, D., Hastie, T. e Tibshirani, An Introduction to Statistical Learning, 2013 (livro-texto), Unofficial Solutions;
* Hastie, T., Tibshirani, R. e Friedman, J., The Elements of Statistical Learning, 2009;
* Tan, Steinbach, and Kumar, Introduction to Data Mining, Addison-Wesley, 2005;
* Bishop, Pattern Recognition and Machine Learning, 2006;
* Ripley, Pattern Recognition and Neural Networks, 1996.
Referência complementar
* Scholkopf and Smola, Learning with Kernels, 2002;
* Mardia, Kent, and Bibby, Multivariate Analysis, 1979;
* Duda, Hart, and Stork, Pattern Classification, 2nd Ed., 2002.
Data Repositories
* Kaggle
* UCI Machine Learning Repository
* Open Gov. Data: dados.gov.br, www.data.gov, www.data.gov.uk;