Monte Carlo Secrets Revealed - A (Zoom) Arena Spectacular
Speaker: Dr Robert Salomone, University of New South Wales
Abstract: Monte Carlo methods, that is, algorithms that employ randomness to estimate some quantity of interest, are widely used in many fields (e.g., Statistics, Machine Learning, Applied Probability, Finance, Econometrics). In this talk, I will give a big-picture overview of some seeming ‘magic’ advanced Monte Carlo techniques that seem like they shouldn’t be possible, highlights include: (i) estimating the probability of something that occurs one time in a trillion without simulating a trillion times (Rare Event Simulation), (ii) obtaining more accurate estimates with the same amount of samples using problem information or even '‘for free'’ (Variance Reduction / Stein Operators), as well as (iii) recent techniques to approximately (but very accurately and quickly) sample from intractable probability distributions, even in the presence of ‘big data’ (Modern Sampling Algorithms) and why this is important. Throughout, I will discuss some examples from my own work, but the talk is intended as an overall conceptual introduction to a variety of advanced Monte Carlo techniques, which will follow a basic introduction.
Biography: Robert Salomone is a Research Fellow with the Australian Centre of Excellence Centre for Mathematical and Statistical Frontiers (ACEMS), based at UNSW Sydney. His research interests primarily lie in computational methodology at the intersection of statistics and machine learning. i.e., using mathematics in smart ways to improve computation for difficult problems. He completed his PhD in Statistics at The University of Queensland in 2018.
Please note for security reasons, you will need to register in advance for this meeting: https://uni-sydney.zoom.us/meeting/register/tJMvdOGqqj4oHt1lENgSdXQacw2HgevZ7KHl
After registering, you will receive a confirmation email containing information about joining the meeting.
Any questions, please feel free to contact: secretary.nswbranch@statsoc.org.au