The Course on Unsupervised Learning in Machine Learning it's yours a gateway to the fascinating world of autonomous machine learning, an area in in full swing with a growing demand for labour. In an environment where Data is the new oil, the ability to to extract value without constant intervention becomes a key skill. This course will equip you to understand and apply advanced techniques such as clustering, dimensionality reduction and generative models, amongst others. You’ll learn how to pre-process data, detect anomalies y to evaluate models efficiently, skills crucial in sectors such as cyber security and finance. With a rigorous theoretical approach y practical examples, you will acquire the the knowledge required to excel in analytical and consultancy roles. Make the most of the opportunity to to train in a discipline that redefines technology and innovation.
Unsupervised Learning in ML
AI Academy
35 hours
Introduction
Objectives
-
Understanding the fundamentals and differences of unsupervised learning.
-
Identify applications of unsupervised learning in industry and technology.
-
Apply pre-processing techniques to improve unsupervised models.
-
Implementing and fine-tuning the K-Means algorithm effectively.
-
Comparing advanced clustering methods and select the most suitable ones.
-
Use PCA and related techniques to reduce dimensionality without any significant loss.
-
Exploring generative models and autoencoders in professional contexts.
Table of Contents
TEACHING UNIT 1. INTRODUCTION TO UNSUPERVISED LEARNING
Fundamentals of machine learning and how it differs from supervised learning
Objectives, principles and characteristics of unsupervised learning
Applications in industry, science and emerging technological fields
Advantages, limitations and operational challenges of unsupervised learning
TEACHING UNIT 2. DATA PRE-PROCESSING FOR UNSUPERVISED MODELS
Normalisation, standardisation and handling of missing data
Variable coding and structural preparation of datasets
Feature selection and basic dimensionality reduction
Best practices for pre-processing in unsupervised algorithms
TEACHING UNIT 3. CLUSTERING: FUNDAMENTALS AND K-MEANS
Key concepts in clustering: distances, centroids and structure
How the K-Means algorithm works internally
Selecting the optimal number of clusters
Conditions for success, limitations and scenarios in which K-Means fails
TEACHING UNIT 4. ADVANCED CLUSTERING METHODS
DBSCAN: density, epsilon, min_samples and use cases
Hierarchical clustering and Mean Shift: principles and advantages
Comparison with K-Means and appropriate parameter selection
Visualisation, interpretation and professional applications
TEACHING UNIT 5. DIMENSIONALITY REDUCTION: PCA AND RELATED TECHNIQUES
Fundamentals of dimensionality reduction and minimal loss of information
PCA: decomposition, principal components and visualisation
t-SNE and UMAP: non-linear structures and high-dimensional representation
Application of these techniques for the analysis and improvement of unsupervised models
TEACHING UNIT 6. GENERATIVE MODELS AND AUTOENCODERS
Key concepts of generative models in unsupervised learning
Autoencoders: architecture, training and variants
Professional applications: compression, reconstruction and anomalies
Generation of synthetic data and its uses in regulated industries
TEACHING UNIT 7. ANOMALY DETECTION
Types of anomalies and their significance in critical environments
Methods based on clustering, distance and density
Autoencoders and statistical approaches to anomalies
Applications in cybersecurity, finance and predictive maintenance
TEACHING UNIT 8. EVALUATION OF UNSUPERVISED MODELS
Challenges of unlabelled evaluation and internal approaches
Key metrics: Silhouette, Davies-Bouldin and others
Indirect assessment, cross-validation and benchmarking
Professional interpretation and decision-making based on unsupervised models
Common professional errors in unsupervised projects and their real-world consequences
TEACHING UNIT 9. PRACTICAL APPLICATIONS AND FINAL PROJECTS
Consultancy-style final project