Upcoming talk events : Nothing yet!
Using Computers to Identify Hurtful speech
Wouldn't it be great if those online bullies just couldn't say those awful things? The media is rife with accounts of children going online and saying hurtful things to each other, adults threatening death or rape to people they have never met before. These words people use to degrade someone else would never be spoken out loud, but behind the computer screen they think it's ok.
I have been working on a side project to use computers and machine learning to identify when someone is typing something hurtful. So far I have managed to create an algorithm that can identify the most likely meaning of a word thus getting a step closer to a computer understanding the meaning behind what you type. I did this using the structure of the english language. Thankfully it is as structured as it is complicated to learn.
My hope is to continue working on this algorithm and get it to work real time, like a spell or grammar checker. Then when running it can stop the offenders being able to post anything that is meant to degrade someone else. It would make online much safer.
The difference between star formation, AGN, and shock emission in merging galaxies
This is really what my PhD is all about. Studying merging galaxies to understand what happens as time goes by in terms of star formation, shocks and AGN processes. What happens when? Why is one process more prevelant at the end of a merge than another? What is going on in the beautiful mess of two galaxies? This has led to a study of optically classified composite galaxies in the CALIFA (Sanchez et al. 2012) and WiGS (Rich et al. 2014) surveys. My research has also included working with the SAMI survey (Croom et al. 2012).
Machine Learning in Astronomy
Large surveys are a huge part of Astronomy. The SKA is an example of one that will be coming online in the next 10 years that will produce more data than we've dealt with before. Whilst working with the SAMI and S7 collaborations I have found that even 100 integral field observations is a lot more than can be easily analysed by just people. I have looked at using an artificial neural network to help us in the specific task of deciding the number of components to be fit to emission lines in IFU surveys. I presented our preliminary results at the ADASS XXV conference in October 2015 along with doing two radio interviews on The Machine (now called LZComp). You can see the media release here: http://www.anu.edu.au/news/all-news/artificial-intelligence-finds-messy-galaxies. Open code for LZComp is now available on GitHub (not for the faint hearted).
Galactic Winds / Outflows in Merging Galaxies
As two galaxies 'collide' gas is funnelled towards the centres of each galaxy, fuelling star formation and central black holes. Later in the merging process the formed stars and black hole can cause massive outflows of material from the centres of the galaxies, revealing previously obscured AGNs and causing Galactic scale winds and shocks. In one galaxy, that I have been studying for my PhD, we have detected an neutral gas outflow using the Na ID (neutral sodium doublet) caused by star formation in a merging galaxy system. The outflow rate is comparable to the star formation rate of the galaxy and matches studies of ionised and molecular gas outflows of this same system. More to come as a paper is written and my PhD is finalised.
(Image of M82; a well known galactic wind galaxy. This is not the galaxy we have been studying, but NASA's image is so much prettier!)
Star formation regions accelerating particles to TeV energies
During my honours years at the University of Adelaide I researched into the star-forming HII region G5.89-0.39 using a Chandra observation. We were looking for evidence of particle acceleration occuring in the star forming region to explain the TeV gamma-ray emission detected by H.E.S.S. You can read about this research in these conference proceedings.