Activity Leaders
Dr Shannon Dillon
- I am a biologist and data scientist, with a background in and passion for genetics. We have been implementing novel methods, including machine learning approaches, to enhance how we integrate multi-layered data streams, and in turn our capacity to deliver new biological understanding and tools for improved productivity of crops to industry. Through CSIROs new MLAI Future Science Platform, we are extending these approaches to other agricultural productions systems, where we will develop and extend machine learning methods to solve biological problems and address key analytical challenges in data driven breeding.
Dr Alex Whan
- I am passionate about leading an activity that will maximise the value of CSIRO’s diverse data through exposing the deep domain knowledge of our scientists to ML/AI methods. In my scientific career I have generated lots of my own data, and analysed lots of scientific data generated by others. This experience has shown me how important data context and description is for analysis, as well as how challenging it is to provide that in an effective and efficient way during generation. Since access to high quality, well described data is essential for most ML/AI applications, enabling scientists to seamlessly their data and how it was made will bridge the gap between domain and analytical experts, and open up new pathways for scientific discovery.
Dr Petra Kuhnert
- The Hybrid Prediction project is aiming to reimagine complex and constrained bio-geo-physical processes with MLAI to support decision-making. I’m excited to be leading a core team of researchers in CSIRO with specialist skills in statistics, computing and machine learning to solve challenging problems in the Australian landscape.
Dr Lars Petersson
- I am excited to be co-leading the MLAI FSP Activity Object Detection as I can see how recent progress in the areas of Computer Vision and Machine Learning can be directly applicable to a large number of real-world problems that several Business Units have, and if addressed will have a real-world impact. For example, it is a common problem that there is abundant access to data, but there are often no labels or other forms of annotations available to train a model that can be used for classification, detection, recognition etc. There are now methods that start addressing such problems, often making use of additional domain specific information (semantics, taxonomies, attributes, physical models etc). In this activity , we will push the research further on methods solving such problems. The types of data that we are mainly concerned with are 2D in nature (images and video), although when available we also make use of additional modalities including, for example, natural language, 3D point clouds, DNA etc.
Dr Vivien Rolland
- A Biologist and Microscopist by training, I lead an A&F research team as well as the Black Mountain microImaging Centre (BMIC). A large part of my work revolves around generating and analysing images, to solve problems across a range of research domains. As such, I am well aware of both the current limitations in image analysis and the potential for MLAI to relieve some of these pain points. One of my current research focus is to explore and develop Machine Learning solutions to quantify microscopic traits that are relevant for Agriculture (Julius career award). In this context, I am very excited to co-lead the Object Detection activity, which seeks at pushing the Machine Learning envelop to deliver practical solutions across a range of domains and across scales.
Dr Peyman Moghadam
- Everything around us can be regarded as a continuum in space and time. This activity is based on breakthrough cross-disciplinary science which aims to develop fundamental and transformative spatio-temporal machine learning/artificial intelligence solutions to characterise complex systems (e.g, agricultural systems, environmental systems, mineral systems or pest outbreaks) and to infer/predict and forecast their changes through space and time.
Dr Iadine Chades
- I develop AI algorithms to provide guidance on how to make smart decisions under imperfect knowledge and resource constraints. I specialise in discovering the mechanistic insights that explain optimal decisions in ecology, health and biosecurity to maximise positive long term impact.
Dr Amir Dezfouli
- I develop machine learning algorithms that can be used in the health domain, with a focus on applications in psychiatry and neuroscience. My research was focused on using deep neural networks and Bayesian methods for developing computational models of the human brain with applications in psychiatry.