Complex Systems

The brain has a remarkable capacity to process information in a fundamentally distributed and dynamical way, outperforming any human-designed computers on our planet. How does such a distributed dynamical computation capacity emerge from the interactions of billions of neurons? What are general physical principles of brain dynamics and functions? The team led by A/Prof. Pulin Gong uses theoretical and modeling approaches, in close collaborations with experimentalists, aims to tackle these questions.

If you’re interested in this research, please contact A/Prof. Pulin Gong.

Deep learning networks widely used in artificial intelligence can be trained to effectively solve many real-world problems such as speech recognition, object detection and drug discovery. We aims to understand why these neural networks in AI work so effectively and develop new types of deep learning algorithms inspired by neural information processing mechanisms.

If you’re interested in this research, please contact A/Prof. Pulin Gong.

Complex systems are all around us, but they often produce surprisingly similar dynamical features. How can we quantify these dynamics efficiently, and discover hidden connections between diverse systems in the world around us? Combining methods from statistical physics, information theory, and machine learning, we have developed new methods to understand the dynamical patterns that emerge from big datasets of complex real-world systems. Our research has been applied to determine the health of ecosystems from streams of audio, to classify different types of stars from their light curves, and to distinguish different psychiatric conditions from brain dynamics.

If you’re interested in this research, please contact Dr Ben Fulcher

Circadian oscillations are present in every physiological function and every living organism on the planet, including plants and insects. We study how these 24-hour rhythms in, e.g., the human sleep-wake cycles, hormone levels, and alertness, are generated and how they are affected by the environmental time cues such as the solar light-dark cycle. We have a particular interest in mechanisms and health consequences of circadian misalignment - desynchronisation of the circadian rhythms from each other and/or from the environment. By using physically-based modelling and data science we quantitatively study this complex biological system, develop predictive tools and design interventions to minimise circadian misalignment.  

If you’re interested in this research, please contact Dr Svetlana Postnova.

Why do we sleep and how does the brain transition into this seemingly unconscious state and out of it? How does sleep loss affect our alertness and memory, and why is long-term sleep loss associated with neurodegenerative disorders like Alzheimer’s? What do sleep and anaesthesia have in common? To answer these and other questions we use neural field, neural mass, and neural network modelling and experimental data collected at different physiological levels – from single neurons to whole brain activity. 

If you’re interested in this research, please contact Dr Svetlana Postnova or Prof. Peter Robinson

Technical advances in recent years have see maps of the brain’s cells and circuits measured in unprecedented detail. We need physics to make sense of the intricate patterns contained in these data: to explain them in terms of physical mechanisms. We use methods from statistics, mathematics, and physics to find, quantify, and explain patterns in the brain, using cell-density data, neural dynamics, gene expression, and axonal connectivity network data. Our ultimate aim is to develop physical, mechanistic accounts of brain organization and function.

If you’re interested in this research, please contact Dr Ben Fulcher

Social media is playing an increasingly important role in our daily lives, spreading information, connecting communities, and ultimately shaping policy positions.  However, traditional social science approaches fail in the face of the vast amounts of data flowing on social media networks.  The methods of data science and the tools of physics allow us to identify the key features of the dynamics, and the models developed to explain these features allow us to better understand why these features emerge.  What factors increase political polarization?  What are the signatures of misinformation flow on the network? What changes in the network in the face of external events, such as the recent bushfires?  To answer these questions we use social network analysis, time-series analysis and agent-based modelling, combined with machine learning techniques and natural language processing.  The findings will help us combat the big social problems of today, such as the increase in political extremism, and the persistence of climate denialism.

If you’re interested in this research, please contact Dr Tristram Alexander