- This event has passed.
ADSN Workshop: Synthetic Data
September 10, 2021 @ 1:00 pm - 3:00 pm
ADSN Workshop: Synthetic Data
We’ve had recent interest from industry about simulating realistic data from complex systems. There are many situations where sensitivities of real data make synthetic data safer to handle. We (QUT) would like to convene a workshop to engage researchers with capabilities and interests in synthetic data and open up possibilities for further connection and collaboration.
Please contact us if you would like to receive an invitation to this online workshop
| 1:00PM – 1:10PM | Welcome & overview of workshop Purpose of workshop and quick introductions |
Distinguish Professor Kerrie Mengersen
QUT Centre for Data Science |
| 1:10PM – 1:20PM | Overview of synthetic data generation
In this overview, Connor will provide a short literature review and summary of main features when it comes to synthetic data generation. |
Conor Hassan QUT Centre for Data Science |
| 1:20PM – 1:35PM | GRATIS: GeneRAting TIme Series with diverse and controllable characteristics
Description: Synthetic time series are useful for benchmarking and testing methods for forecasting, clustering, classification and other tasks. I will discuss an approach to this where we can generate time series with diverse and controllable characteristics using mixture autoregressive (MAR) models. This can be done with the gratis package for R. |
Professor Rob Hyndman Monash University |
| 1:35PM – 1:50PM | Deep Learning Techniques for Dealing with Lack of Data In this talk, we will present progressive transfer learning methods to deal the data problems such as the lack of labelled data and data shifts. We present an alternative approach to complement the datasets available. |
Associate Professor Richi Nayak QUT Centre for Data Science |
|
Opportunity for Spotlight Talks |
||
| 2:00PM – 2:15PM | Synthetic data generation using moment-based density estimation
In this talk, we introduce a new synthetic data generation method based on estimated multivariate density function, which is constructed from the sample moments information of the original data. |
Bradley Wakefield University of Wollongong (NIASRA) |
| 2:15PM – 2:30 PM | Generating artificial video data to train machine learning algorithms This talk will present early work on the creation of synthetic video data using motion capture and CGI for use in training of human action recognition models. |
Anthony Paproki QUT Centre for Data Science |
|
Opportunity for discussion |
||