Statistics Across Campuses

  On behalf of SSA NSW branch, we would like to invite you to join the virtual event we are holding on Wednesday September 21st at 6.30 pm. Tom Honeyman from   Australian Research Data Commons will speak about  recognition of Research Software as a first-class output of research. Date:  Wednesday, 21st September Time:   6.30 pm Location:  This is a virtual event. You will need to register in advance  here  to obtain the Zoom link.  Speaker:   Tom Honeyman Title:  Recognising the value of research software Abstract: The ARDC has a program of activities to work towards recognition of Research Software as a first-class output of research. We want to change the prevailing culture around the invisible status of what we consider to be a valuable and uniquely actionable form of knowledge representation. In this talk, I'll cover the actions we're undertaking to make software a more visible and recognised undertaking in Research, and I'll cover what might be ways for you to make you

  Why interpolating neural nets generalize well: recent insights from neural tangent model Date :  Thursday, 22 September 2022 Time:   10:00  - 11:00 am Speaker:  Dr J oe (Yiqiao) Zhong  ( University of Wisconsin-Madison ) Abstract: A mystery of modern neural networks is their surprising generalization power in overparametrized regime: they comprise so many parameters that they can interpolate the training set, even if actual labels are replaced by purely random ones; despite this, they achieve good prediction error on unseen data. In this talk, we focus on the neural tangent (NT) model for two-layer neural networks, which is a simplified model. Under the isotropic input data, we first show that interpolation phase transition is around Nd ~ n, where Nd is the number of parameters and n is the sample size. To demystify the generalization puzzle, we consider the min-norm interpolator and show that its test error/generalization error is largely determined by a smooth, low-degree component

    Data Science and Imaging the Black Hole Shadow. Date: Tuesday, 13 September 2022   Time: 6:00pm (AEST)   Speaker:  Prof. Shiro Ikeda (Institute of Statistical Mathematics)   Contact the organizer: Andriy Olenko a.olenko@latrobe.edu.au Abstract: In April 2019, the EHTC (Event Horizon Telescope collaboration) released the first image of the M87 black hole shadow and this May, the black hole shadow image of our Milky Way galaxy was released. The EHTC has more than 300 members from different backgrounds and countries. I have been involved in this project as a data scientist for more than 8 years and collaborated with EHTC members to develop a new imaging method. The EHT is a huge VLBI (very long baseline interferometer), which is different from optical telescopes in that a lot of computation is required to obtain a single image. Black hole imaging is also very interesting from the data scientific viewpoint. In this talk, I will explain how the new imaging technique has been developed

Advanced Bayesian approaches for state-space models with a case study on soil carbon sequestration Speaker: Mohammad Davoudabadi Zoom: https://uni-sydney.zoom.us/j/85328587917 Abstract: Sequestering carbon into the soil can mitigate the atmospheric concentration of greenhouse gases, improving crop productivity and yield financial gains for farmers through the sale of carbon credits. In this work, we develop and evaluate advanced Bayesian methods for modelling soil carbon  sequestration and quantifying uncertainty around predictions that are needed to fit more complex soil carbon models, such as multiple-pool soil carbon dynamic models. This study demonstrates efficient computational methods using a one-pool model of the soil carbon dynamics previously used to predict soil carbon stock change under different agricultural practices applied at Tarlee, South Australia. We focus on methods that can improve the speed of computation when estimating parameters and model state variables in a st

Dynamic Change Detection with Application to Skilled Funds Selection Date :  Thursday, 8 September 2022 Time:   4:00  - 5:00 pm Speaker:     Dr Lilun Du  (Hong Kong University of Science and Technology) Abstract: Nowadays, data often arrive in large-scale streams, making timely and accurate decision necessary. It has become important to rapidly and sequentially identify individual stream whose behavior deviates from the norm in the framework of false discovery rate control. By fully exploiting the sequential feature of data streams, we introduce a distribution-free procedure which imposes minimum requirement on the knowledge of streaming observations. The idea is to combine an order-preserved sample-splitting strategy with exponentially weighted moving average approach to construct a series of statistics with marginal symmetry property, and then to utilize the symmetry property for obtaining a data-driven threshold. Finite-sample and asymptotic results on the false discovery rate contr