Computational Methods for Hidden Dynamics in Time Series - Presented by Joonha Park

Many real-world systems—such as the spread of infectious diseases, ecological populations, or financial markets—evolve over time in ways that are only partially observable. How can we learn about the underlying mechanisms driving these processes?

In this talk, I will introduce the idea of hidden Markov models, a mathematical framework for describing systems whose internal states are not directly observed. I will discuss computational tools, including the Kalman filter and Monte Carlo methods, that allow us to estimate hidden states and model parameters from data.

Through applied examples, including a compartmental model for infectious disease dynamics, I will demonstrate how modern computational techniques make it possible to analyze complex nonlinear time series models.