Mediaspace scheduled maintenance: Aug 25, 2026 07:00 - 12:00 AM. During this time, videos will be temporarily unavailable. Check status updates.

Lecture

Variable Selection Methods: Subset vs. Container Approaches

In course

EE-311: Fundamentals of machine learning

Ce cours présente une vue générale des techniques d'apprentissage automatique, passant en revue les algorithmes, le formalisme théorique et les protocoles expérimentaux. ...

Description

This lecture covers the concept of variable selection in machine learning, focusing on subset selection and container methods. Subset selection involves determining the best subset of variables for a learning algorithm, while container methods use algorithms to select variables. The instructor explains the naive method, which exhaustively tries all combinations of variables, and non-exhaustive methods like forward and backward search. The lecture also introduces floating search, a greedy procedure for selecting variables. Examples and workout scenarios are provided to illustrate the different selection approaches.

This video is available exclusively on Mediaspace for a restricted audience. Please log in to MediaSpace to access it if you have the necessary permissions.

Watch on Mediaspace

Instructors (2)

François Fleuret

Michael Stefan Daniel Liebling

Official source

Related concepts (24)

Feature selection

Feature selection is the process of selecting a subset of relevant features (variables, predictors) for use in model construction. Stylometry and DNA microarray analysis are two cases where feature selection is used. It should be distinguished from feature extraction. Feature selection techniques are used for several reasons: simplification of models to make them easier to interpret by researchers/users, shorter training times, to avoid the curse of dimensionality, improve data's compatibility with a learning model class, encode inherent symmetries present in the input space.

Directional selection

In population genetics, directional selection, is a mode of negative natural selection in which an extreme phenotype is favored over other phenotypes, causing the allele frequency to shift over time in the direction of that phenotype. Under directional selection, the advantageous allele increases as a consequence of differences in survival and reproduction among different phenotypes. The increases are independent of the dominance of the allele, and even if the allele is recessive, it will eventually become fixed.

Disruptive selection

Disruptive selection, also called diversifying selection, describes changes in population genetics in which extreme values for a trait are favored over intermediate values. In this case, the variance of the trait increases and the population is divided into two distinct groups. In this more individuals acquire peripheral character value at both ends of the distribution curve. Natural selection is known to be one of the most important biological processes behind evolution.

Natural selection

Natural selection is the differential survival and reproduction of individuals due to differences in phenotype. It is a key mechanism of evolution, the change in the heritable traits characteristic of a population over generations. Charles Darwin popularised the term "natural selection", contrasting it with artificial selection, which is intentional, whereas natural selection is not. Variation exists within all populations of organisms. This occurs partly because random mutations arise in the genome of an individual organism, and their offspring can inherit such mutations.

Frequency-dependent selection

Frequency-dependent selection is an evolutionary process by which the fitness of a phenotype or genotype depends on the phenotype or genotype composition of a given population. In positive frequency-dependent selection, the fitness of a phenotype or genotype increases as it becomes more common. In negative frequency-dependent selection, the fitness of a phenotype or genotype decreases as it becomes more common. This is an example of balancing selection.

Related lectures (32)

Specification Testing and Machine Learning

Explores specification testing, machine learning, overfitting, regularization, prediction tests, and variable selection.

Linear Regression: Statistical Inference and Regularization

Covers the probabilistic model for linear regression and the importance of regularization techniques.

Discrete choice and machine learning: two methodologies MOOC: Selected Topics on Discrete Choice

Delves into the complementary methodologies of discrete choice and machine learning, covering notations, variables, models, data processes, extrapolation, what-if analysis, and more.

LASSO Regression: Sparse Signal Induction

Explores LASSO regression for inducing sparsity in signals through gradient descent.

Discrete Choice Analysis