Course

MICRO-455: Machine learning I

Lectures in this course (60)

Covers sampling, cross-validation, quantifying performance, optimal model determination, overfitting detection, and classification sensitivity.

SVM - Principle: Linear Classifiers

Covers the history and applications of SVM, as well as the construction of linear classifiers and the concept of classifier margin.

Linear SVM derivation

Covers the derivation of Linear Support Vector Machine (SVM) and the Karush-Kuhn-Tucker (KKT) conditions.

SVM for Non-separable Datasets

Explains SVM for non-separable datasets, introducing slack variables and optimizing the margin for classification.

Quantifying Performance: Misclassification and F-Measure

Covers quantifying performance through true positives, false negatives, and false positives in machine learning.

Nonlinear SVM: Kernels and Dual Optimization

Explores transforming data with nonlinear maps, kernels, dual optimization, and interpreting SVM results.

SVM Hyperparameters

Delves into SVM hyperparameters, showcasing the impact of C and kernel width on classification results.

Multiclass SVM

Covers the use of Support Vector Machines for multi-class classification and the importance of support vectors in tightening classification boundaries.

SVM: Generalization & Confidence

Explores SVM's generalization, lack of confidence in predictions, and risk of false positives, comparing it with GMM and KNN.

Support Vector Machine Overview

Gives an overview of Support Vector Machines, comparing advantages and disadvantages of SVM with other classifiers.

Neural Networks: Perceptron

Covers the main concepts of neural networks, including the Perceptron model and training algorithms.

Neural Networks: Multi-layers

Explains the learning process in multi-layer neural networks, including back-propagation, activation functions, weights update, and error backpropagation.

Deep Neural Networks: Training and Optimization

Explores deep neural network training, optimization, preventing overfitting, and different network architectures.

Nonlinear Regression: Forecasting Trends

Explores machine learning techniques for nonlinear regression and forecasting trends in complex data sets.

Linear & Weighted Regression

Covers linear and weighted regression, limitations, and introduces locally weighted regression for model-free regression.

Ridge Regression: Regularizing Linear Models

Introduces ridge regression to regularize linear models and the kernel trick for nonlinear regression.

Support Vector Regression: Principles and Optimization

Covers Support Vector Regression principles, optimization, and hyperparameters' influence on the fit.

Gaussian Mixture Regression: Modeling and Prediction

Covers Gaussian Mixture Regression principles, modeling joint and conditional densities for multimodal datasets.

Comparison Across Methods: GMR vs SVR

Compares GMR and SVR in machine learning, discussing their similarities, differences, and hyperparameters.

Kernel PCA: Nonlinear Dimensionality Reduction

Explores Kernel Principal Component Analysis, a nonlinear method using kernels for linear problem solving and dimensionality reduction.