Unsupervised Learning & Time Series

Step 2 Stochastic Dynamics Probability 2.5 Stochastic Process & Time Series

A generic course about Stochastic Process & Time Series. Content coming soon.

Self-Supervised Learning Theory

Theory of contrastive and predictive self-supervised learning: objectives, augmentations, and representation quality.

Latent Variable Models: Identifiability & EM Theory

Identifiability of latent variable models, EM algorithm convergence, and modern extensions.

Probabilistic Forecasting

Beyond point forecasts: distributional, quantile, and conformal prediction for time series.

State-Space Models & Kalman Variants

Linear and nonlinear state-space models with Kalman filtering and modern deep SSM extensions.

Hidden Markov Models & State-Space Theory

HMM theory: forward-backward algorithm, Baum-Welch, and spectral learning methods.

Kalman & Particle Filters

Bayesian filtering: from Kalman filters to sequential Monte Carlo for nonlinear/non-Gaussian systems.

Spatiotemporal Processes & Gaussian Random Fields

Gaussian random fields, kriging, and spatiotemporal covariance modeling for environmental and scientific data.

Spatiotemporal Modeling (Graph Time Series)

Graph-based approaches to spatiotemporal forecasting using GNNs and temporal graph networks.

Hierarchical & Intermittent Demand Forecasts

Forecast reconciliation for hierarchical time series and methods for intermittent/sparse demand.

Demand Sensing & Causal Nowcasting

Real-time demand estimation combining causal methods with high-frequency signals for nowcasting.

Anomaly Detection & Change-Point Methods

Statistical methods for detecting anomalies and change points in time series data.