Advanced Machine Learning: Feature Selection

This lecture covers the fundamental components of machine learning algorithms, including input description, internal representation, and decision mechanisms. It delves into feature selection techniques such as FAST and BRIEF descriptors, emphasizing the importance of selecting relevant features in large datasets. The instructor also discusses the limitations of deep learning, highlighting issues like algorithmic bias, lack of diversity, and susceptibility to deception, urging the audience to explore a variety of machine learning techniques beyond deep learning for a more comprehensive understanding of the field.

About this result

This page is automatically generated and may contain information that is not correct, complete, up-to-date, or relevant to your search query. The same applies to every other page on this website. Please make sure to verify the information with EPFL's official sources.

Related lectures (38)

Related concepts (32)

Alpha diversity

In ecology, alpha diversity (α-diversity) is the mean species diversity in a site at a local scale. The term was introduced by R. H. Whittaker together with the terms beta diversity (β-diversity) and gamma diversity (γ-diversity). Whittaker's idea was that the total species diversity in a landscape (gamma diversity) is determined by two different things, the mean species diversity in sites at a more local scale (alpha diversity) and the differentiation among those sites (beta diversity).

Beta diversity

In ecology, beta diversity (β-diversity or true beta diversity) is the ratio between regional and local species diversity. The term was introduced by R. H. Whittaker together with the terms alpha diversity (α-diversity) and gamma diversity (γ-diversity). The idea was that the total species diversity in a landscape (γ) is determined by two different things: the mean species diversity at the local level (α) and the differentiation among local sites (β).

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.

Diversity index

A diversity index is a quantitative measure that reflects how many different types (such as species) there are in a dataset (a community), and that can simultaneously take into account the phylogenetic relations among the individuals distributed among those types, such as richness, divergence or evenness. These indices are statistical representations of biodiversity in different aspects (richness, evenness, and dominance).

Large language model

A large language model (LLM) is a language model characterized by its large size. Their size is enabled by AI accelerators, which are able to process vast amounts of text data, mostly scraped from the Internet. The artificial neural networks which are built can contain from tens of millions and up to billions of weights and are (pre-)trained using self-supervised learning and semi-supervised learning. Transformer architecture contributed to faster training.

Real-world engineering applications must cope with a large dataset of dynamic variables, which cannot be well approximated by classical or deterministic models. This course gives an overview of method

Learning Sparse Features: Overfitting in Neural Networks PHYS-754: Lecture series on scientific machine learning

Discusses how learning sparse features can lead to overfitting in neural networks despite empirical evidence of generalization.

Landscape and Generalisation in Deep Learning

Explores the challenges and insights of deep learning, focusing on loss landscape, generalization, and feature learning.

Feature Engineering: Polynomial Regression

Covers fitting linear regression on features of the original predictors for flexible feature representation.

Machine Learning: Features and Model Selection CS-401: Applied data analysis

Delves into the significance of features, model evolution, labeling challenges, and model selection in machine learning.

Feature Learning: Stability and Curse of Dimensionality PHYS-754: Lecture series on scientific machine learning

Explores how modern architectures beat the curse of dimensionality and the importance of stability in deep learning models.