AUT's departments and institutes are working on a variety of projects, ranging from autonomous vehicles over intelligent surveillance to brain-inspired AI.
The NeuCube is a brain-inspired Spiking Neural Network architecture that can be used to analyse spatio-temporal data. It was developed by Professor Nikola Kasabov at KEDRI and has since been successfully applied to a variety of application areas.
The Brain Data Analysis Lab was established in November 2015. Its main objectives are to offer contemporary tools and systems for brain data modelling, analysis and understanding, including EEG data, fMRI data, and integrated static and dynamic brain data.
Its applications span across Neuromarketing, EEG and fMRI data modelling and Brain computer interfaces (BCI).
The Bio- and Health-Informatics Lab is developing new methods and systems for biological and health data analysis, modelling, prediction and understanding. The Lab is working on several projects related to personalised modeling.
The aim of this project is to model and analyse real-time environmental data at a large scale. This includes predictive modeling of seismic data, air pollution data modeling, and wind turbine energy prediction.
For example, we use Spiking Neural Networks to predict earthquakes by identifying seismic P- and S-waves. Quickly detecting and accurately identifying the first arriving P-waves is important in earthquake forecasting as they contain information about ground movements that might lead to seismic activity. We collect and label seismic data and study the features of these data to learn about their characteristics. The outcomes of our P-wave detection method will be compared with other deep learning methods for time series analysis.
The Signal Processing & Pattern Recognition Lab is developing new methods and systems for signal analysis, pattern recognition, pattern understanding and predictive data modelling from from image and video data. The lab is working on Fast Moving Object Recognition and Age Invariant Face Recognition.
Jesús López Lobo (Txus), as Visiting Researcher of the PANTHER programme from November 2017 to April 2018, focused on adapting the Evolving Spiking Neural Networks (eSNNs) to real data stream scenarios and specifically worked on online learning and concept drift. Data stream scenarios are characterised by huge amounts of data that flow fast and continuously, and often impose a set of memory use and runtime processing restrictions that any learning algorithm should adequately tackle. Under these circumstances, Txus has adapted eSNNs to be a competitive algorithm for concept drift adaptation, and he has also provided KEDRI with a drift detector based on an eSNN, both of them fulfilling the online learning requirements.
The Study of Creativity Lab is developing new methods and systems for audio-visual information processing, for concept formation, learning languages and Virtual Reality visualisation for the data and process understanding.
This is a collaboration between KEDRI, AUT CoLab, and AUT Sentience Lab.
The overall objective of this research project is to enable brick and mortar (BaM) retailers to track what their customers are looking at, so that they can react to that information immediately and can better withstand the pressures of online retail. This gives BaM retailers unprecedented insight into optimal sales processes, as well as a chance to make immediate personalized offers in a way that online retailers do.
The project will analyse ESR's forensic casework data to identify patterns in the types, timing and outputs of the forensic science conducted. Through this analysis Forensic Intelligence outputs will be produced identifying the forensic examinations that will help to improve the way New Zealand prevents, responses and supports the recovery from crime.
The project aim is in the formulation of temporal taste profiles for microblog users based on the content they disseminate and homophily in their social network structure. The research process encompasses extraction of personalized temporal user interests in microblog texts and formulation of fitting profiles. This research is fundamental in services related to content recommendations and audience measurement.
Research in sports science enables automated coaching experiences for end-users to improve their motion control, skills, and technique by using the next generation of autonomous intelligent Augmented Coaching Systems (AGS), exergames, and virtual or immersive environments. An essential component of coaching, injury recovery and (re)learning a new motion/skill/technique is the inherent ability of AGS to measure motion data and perform qualitative assessments based on subjective criteria by combining traditional algorithmic and machine learning paradigms. A nature-inspired machine implementation of subjective and adaptive criteria found in qualitative assessment of human motion relates to a coach's implicit insights, personal knowledge of a subject and evolving improvement goals.
The Centre for Robotics and Vision develops systems that employ machine learning and AI techniques for analysing image and video data. Projects include navigation and driver assistance for autonomous vehicles and unmanned aerial vehicles (UAVs), as well as intelligent surveillance systems.
Our mission is to create, develop and commercialise innovative IT products. The Centre’s primary goal is to provide a stimulating environment for the development of these different forms of technology in New Zealand under strong leadership. Our research is not ivory-towered. Given the wide range of research in IT, the Centre has decided to focus on four main areas of research and development - human language technology, speech technology, robotics and mind theory.
The Data Science Research Group (DSRG) seeks to promote research into all aspects of Data Science. Data Science is poised to become one of the most intensively researched areas within the Computer Science discipline and the research carried out by the group will be at the cutting edge of new developments in this field.