Unsupervised Learning — Clustering, Manifold & Topic Models — Hands-on

Learn how to explore high dimensional biomedical and omics data without labels using clustering, manifold learning and topic models. This module walks through similarity measures, clustering families, dimensionality reduction and topic modeling, with a strong focus on diagnostics, stability and clear reporting in R and Python.

Unsupervised Learning — Clustering, Manifold & Topic Models

Help Desk · WhatsApp

Session Index

Session 1 — Unsupervised Learning Foundations & Similarity Session 2 — Clustering Algorithms & Cluster Evaluation Session 3 — Manifold Learning & Dimensionality Reduction Session 4 — Topic Models & Unsupervised Patterns in Omics

Session 1

Fee: Rs 8800

Unsupervised Learning Foundations & Similarity

When and why to use unsupervised learning

exploratory structure discovery subtyping and endotypes feature space thinking

Distance and similarity measures

Euclidean, Manhattan, cosine correlation based distances choice of metric vs data type

Preprocessing for unsupervised workflows

scaling and transformation handling outliers and noise feature selection and filtering

Session 2

Fee: Rs 11800

Clustering Algorithms & Cluster Evaluation

Partitioning and hierarchical clustering

k means and k medoids agglomerative hierarchical clustering linkage choices and dendrograms

Density and model based clustering

DBSCAN and HDBSCAN Gaussian mixture models soft vs hard assignments

Cluster quality and stability

silhouette and Davies Bouldin internal vs external validation resampling based stability checks

Session 3

Fee: Rs 14800

Manifold Learning & Dimensionality Reduction

Linear methods

PCA and variance explained PCA biplots and loadings NMF for parts based patterns

Nonlinear embeddings

t SNE concepts and pitfalls UMAP for neighborhood preservation hyperparameter sensitivity

Using embeddings in downstream analyses

visualization of clusters and batches feeding embeddings into ML models reproducible seed management

Session 4

Fee: Rs 18800

Topic Models & Unsupervised Patterns in Omics

Topic modeling concepts

bag of words intuition LDA style topic models document topic and word topic matrices

Extensions to omics and clinical features

using topics as molecular programs interpreting topic loadings linking topics to outcomes

Deliverables: unsupervised analysis report

cluster and embedding plots topic tables and summaries R and Python scripts or notebooks

PDF

Unsupervised Learning — Clustering, Manifold & Topic Models Training | Biostatistics & ML for Omics

Unsupervised Learning — Clustering, Manifold & Topic Models — Hands-on