Swinburne Youth Innovation Challenge
Swinburne started the space challenged to introduce high school students to the amazing world of space science and industry. This unique 11 week program combines lectures, tutorials, team work and creativity for our students. Students are also part of a real-life space experiment heading to the ISS. Open to years 10-12 around Australia.
Artificial Intelligence as the Most Valuable Player: Enabling cyber-human teams to achieve decision superiority
Current Postdoctoral Research
PI: Prof. Christopher Fluke and A.Prof. Clare MacMahon
Human performance in time-critical decision making is influenced by a combination of mental workload, stress, situational awareness and expertise. Decision-making processes in high-risk, time-pressured scenarios are now regularly supported by artificial intelligence (AI), however, the integration of computational support through the inclusion of one or more AI team members needs to be approached thoughtfully. This can be achieved by designing intelligent agents that are not only human-centred but are team-centred. The aims of this project are: (1) to investigate the role of an AI team-member as a monitor of individual decision-making performance; (2) to study the characteristics of effective cyber-human teams through the integration of an AI team-member; and (3) to explore the role of an AI team-member in human-machine teams where the AI is the MVP.
Swinburne University of Technology and La Trobe University
Human performance in time-critical decision making is influenced by a combination of mental workload, stress, situational awareness and expertise. Decision-making processes in high-risk, time-pressured scenarios are now regularly supported by artificial intelligence (AI), however, the integration of computational support through the inclusion of one or more AI team members needs to be approached thoughtfully. This can be achieved by designing intelligent agents that are not only human-centred but are team-centred. The aims of this project are: (1) to investigate the role of an AI team-member as a monitor of individual decision-making performance; (2) to study the characteristics of effective cyber-human teams through the integration of an AI team-member; and (3) to explore the role of an AI team-member in human-machine teams where the AI is the MVP.
Swinburne University of Technology and La Trobe University
The Deeper Wider Faster program (Co-PI)
PI: Prof. Jeff Cooke
The night sky is more dynamic than one can even imagine, from the luminous and energetic deaths of stars, to the violent mergers of blackholes.
The Deeper, Wider, Faster program (DWF) chases the fastest bursts in the sky with multi-facility, multi-wavelength, simultaneous observations and rapid follow up. DWF probes the milliseconds-to-hours time domain with fast-cadence, deep observations of wide regions of sky. The most elusive transients that DWF aims to discover and study include counterparts to fast radio bursts and gravitational wave events.
The DWF program is not a 'reactive' follow up program, like all programs previously, our approach is "proactive". Target fields are observed simultaneously, and continuously with multiple telescopes, allowing us to gather critical data before, during, and after the fast event. If a transient such as a fast radio burst is detected during DWF observations, several telescopes are already observing that part of the sky. We process the data in real-time on the Swinburne supercomputer OzSTAR and are able to identify and confirm detections within minutes of the outburst with software and our team of researchers at Swinburne and around the world. The fast identifications are important to trigger follow-up deep spectroscopy and imaging minutes later using the world's largest telescopes on the ground and telescopes in space.
Work as Co-PI during PhD:
The night sky is more dynamic than one can even imagine, from the luminous and energetic deaths of stars, to the violent mergers of blackholes.
The Deeper, Wider, Faster program (DWF) chases the fastest bursts in the sky with multi-facility, multi-wavelength, simultaneous observations and rapid follow up. DWF probes the milliseconds-to-hours time domain with fast-cadence, deep observations of wide regions of sky. The most elusive transients that DWF aims to discover and study include counterparts to fast radio bursts and gravitational wave events.
The DWF program is not a 'reactive' follow up program, like all programs previously, our approach is "proactive". Target fields are observed simultaneously, and continuously with multiple telescopes, allowing us to gather critical data before, during, and after the fast event. If a transient such as a fast radio burst is detected during DWF observations, several telescopes are already observing that part of the sky. We process the data in real-time on the Swinburne supercomputer OzSTAR and are able to identify and confirm detections within minutes of the outburst with software and our team of researchers at Swinburne and around the world. The fast identifications are important to trigger follow-up deep spectroscopy and imaging minutes later using the world's largest telescopes on the ground and telescopes in space.
Work as Co-PI during PhD:
- Lead of systematic archival data reduction pipelines for DECam data (15,000+ 3deg^2 images!)
- Creation of data products ready for transient discovery and science, eg deep stacks, light curves
- Leading real-time data transfer/reduction pipelines
- Communicating transient events among several facilities in real-time for follow-up/targeted observations.
- Leading telescope proposals for Australian optical facilities
- Supervision of work experience high school students
|
|
Unsupervised Machine learning for Lightcurves
A key part of my work for DWF has involved the creation of millions of fast cadenced (~60sec) lightcurves from the archival data collected during DWF. For my main science goals we needed to be able to identify anomalous events within in the light cruves, but not limited ourselves to only known events, biasing our results. To solve this, I worked closely with Dr. Michelle Lochner during the Kavli Summer program in Astrophysics, and we explored using unsupervised clustering to isolate the different types of lightcurves within our data. These anomalies were then ranked using an isolation forest built into the python package Astronomaly.
We tested these methods originally on a set of ~85,000 DWF light curves and successfully identified all known variables and discovered previously uncatalogued variable stars and flaring stellar sources. See our work here.
We tested these methods originally on a set of ~85,000 DWF light curves and successfully identified all known variables and discovered previously uncatalogued variable stars and flaring stellar sources. See our work here.
|
|
SHINE Microgravity experiment - Tooth Decay
Overview: From 2018 on i'veI had the pleasure of working as a mentor on the Swinburne Haileybury International Space Station Experiment (SHINE). During this time we've worked towards research on tooth decay in Microgravity. The aptly named experiment "MircoCavity" was designed and built by 6 talented senior students of Haileybury High School, supervised by a group of senior mentors including Astrophysicists, student Engineers and microbiologist. Being part of this project is an incredible experience and taught us how to problem solve in the most creative and abstract ways. The tooth decay project remained on the ISS for a total of 30 days and was completely automated by software written by our talented students. Our
The specifics: Our 12x5x5 centimetre experiment may of been little but it was mighty! Insight housed a custom designed 3D printed experiment chamber containing a real human tooth, tooth like substrate and freeze dried streptococcus mutans. Connected to this was a fluid sack containing activation broth connected to the chamber with a mirco pump.
The specifics: Our 12x5x5 centimetre experiment may of been little but it was mighty! Insight housed a custom designed 3D printed experiment chamber containing a real human tooth, tooth like substrate and freeze dried streptococcus mutans. Connected to this was a fluid sack containing activation broth connected to the chamber with a mirco pump.
|
|
Australian Dark Energy Survey
Overview: "As the leading source of spectroscopy for the Dark Energy Survey, the OzDES team used the Anglo-Australian Telescope during a six year observing program designed to measure the redshifts of tens of thousands of galaxies and obtain spectra of supernovae and other transients. The galaxy redshifts will be used to make the most detailed measurement of the Universe's expansion history ever and will lead to a better understanding of the physics behind the acceleration of the Universe." Read more on OzDES HERE.
The specifics: I worked as part of OzDES/DES team during my honours degree, specifically researching contamination effects of Core-Collapse Supernovae within the photometric DES pipelines. This work including simulating the 3 year DES observations on the CTIO-DECam instrument with volumetrically calculated Supernovae rates. Through this approach we were able to determine that 1.24% percentage of photometrically classified type Ia Supernovae would of originated from a Core-Collapse supernovae event. With this level of contamination, we find a 0.5% error in the fitted modelling of the dark energy equation of state, and 1.5% error in estimating the matter density of the universe. Full thesis bellow.
The specifics: I worked as part of OzDES/DES team during my honours degree, specifically researching contamination effects of Core-Collapse Supernovae within the photometric DES pipelines. This work including simulating the 3 year DES observations on the CTIO-DECam instrument with volumetrically calculated Supernovae rates. Through this approach we were able to determine that 1.24% percentage of photometrically classified type Ia Supernovae would of originated from a Core-Collapse supernovae event. With this level of contamination, we find a 0.5% error in the fitted modelling of the dark energy equation of state, and 1.5% error in estimating the matter density of the universe. Full thesis bellow.
