Kept on Wikipedia:Australian Artificial Intelligence Institute

---

Australian Artificial Intelligence Institute (AAII)

Type	Research institute
Established	2017
Affiliation	University of Technology, Sydney
Director	Associate Professor Jie Lu AO
Location	Sydney , Australia
People	200+ scientific researchers and professional staff
Website	www.uts.edu.au/research/australian-artificial-intelligence-institute

The Australian Artificial Intelligence Institute (AAII) is a research institute in artificial intelligence hosted by the University of Technology, Sydney.

AAII’s aim is to develop theoretical foundations and advanced algorithms for artificial intelligence, and the related areas of computational intelligence, computer vision, network analysis, data science, machine learning, pattern recognition, bioinformatics, brain-computer interface, social robotics, and intelligent information systems.

Objectives

1. To construct theoretical foundations, frameworks and methodologies for AI.
2. To develop advanced technological capabilities of AI, including translatable models, tools, prototypes and systems.
3. To explore transformative applications with industry and government partners for AI.
4. To establish strong university-industry-government and international-domestic partnerships for AI research.
5. To train the next generation of researchers and scientists in AI.

History

In 2007, a collaboration of four main laboratories in intelligent systems became the Centre for Intelligent Information Systems (CIIS).

In 2008, with the addition of a new Quantum Computation laboratory, CIIS became the Centre for Quantum Computation & Intelligent Systems (QCIS).

The Centre for Artificial Intelligence (CAI), which was officially launched in March 2017, merging four AI-related labs from QCIS and the Computational Intelligence and Brain Computer Interface (CIBCI).

It was renamed the Australian Artificial Intelligence Institute in August 2020.

AAII currently has 35 core member academic staff, over 200 PhD students and 8 research labs.

Research

The Australian Artificial Intelligence Institute researches across a wide range of areas in artificial intelligence.

Fundamental research areas:

• Computational Intelligence
• Deep Learning
• Transfer Learning
• Large-scale Graph Processing
• Concept Drift
• Reinforcement Learning
• Pattern recognition
• Probabilistic Machine Learning
• Big Dimensionality
• Neuromorphic Computing
• AI-Driven Software Security Analysis
• Computer Vision
• Explainable AI

Technology and transfer research:

• Brain Computer Interface
• Recommendation Systems
• Social Networks
• Social Robotics
• Decision Support Systems
• Cloud Computation
• Blockchain
• Human Autonomy team
• Bioinformatics
• Data Science and Visualisation
• Text Mining
• AI Privacy & Security
• Network Analytics

AI Technologies

AAII staff have published over 1300 papers, with over 450 being in in ERA/CORE A*/A, Q1 journals or Core Tier A Conferences.

These include:

• IEEE Transactions on Pattern Analysis and Machine Intelligence (IEEE TPAMI)
• IEEE Transactions on Neural Networks and Learning Systems (IEEE TNNLS)
• IEEE Transactions on Cybernetics (IEEE TCYB)
• IEEE Transactions on Fuzzy Systems (IEEE TFS)

Some selected technologies developed by AAII researchers include:

RFNN: Robust Fuzzy Neural Network With an Adaptive Inference Engine

Robust Fuzzy Neural Network (RFNN) contains an adaptive inference engine that is capable of handling samples with high-level uncertainty and high dimensions. Unlike traditional FNNs that use a fuzzy AND operation to calculate the firing strength for each rule, RFNN is able to learn the firing strength adaptively. It also further processes the uncertainty in membership function values. Taking advantage of the learning ability of neural networks, the acquired fuzzy sets can be learned from training inputs automatically to cover the input space well. Furthermore, the consequent layer uses neural network structures to enhance the reasoning ability of the fuzzy rules when dealing with complex inputs.

SSD: Multidomain Adaptation with Sample and Source Distillation

Multidomain adaptation method with sample and source distillation (SSD) develops a two-step selective strategy to distill source samples and define the importance of source domains. To distill samples, the pseudo-labeled target domain is constructed to learn a series of category classifiers to identify transfer and inefficient source samples. To rank domains, the agreements of accepting a target sample as the insider of source domains are estimated by constructing a domain discriminator based on selected transfer source samples. Furthermore, an enhancement mechanism is built by matching selected pseudo-labeled and unlabeled target samples. The degrees of acceptance learned by the domain discriminator are finally employed as source merging weights to predict the target task.

DCA: Dynamic Classifier Alignment for Unsupervised Multi-Source Domain Adaptation

A dynamic classifier alignment (DCA) method for multi-source domain adaptation, which aligns classifiers driven from multi-view features via a sample-wise automatic way. To determine the important degrees of multiple views, an importance learning function is built by generating an auxiliary classifier. To learn the source combination parameters, a domain discriminator is developed to estimate the probability of a sample belonging to multiple source domains. Meanwhile, a self-training strategy is proposed to enhance the cross-domain ability of source classifiers with the assistance of pseudo target labels.

MSCLDA: Multi-source contribution learning for domain adaptation

A novel multi-source contribution learning method for domain adaptation (MSCLDA). As proposed, the similarities and diversities of domains are learned simultaneously by extracting multi-view features. Then multi-level distribution matching is employed to improve the transferability of latent features, aiming to reduce misclassification of boundary samples by maximizing discrepancy between different classes and minimizing discrepancy between the same classes. Concurrently, instead of averaging source predictions or weighting sources using normalized similarities, the original weights learned by normalizing similarities are adjusted using pseudo target labels to increase the disparities of weight values, which is desired to improve the performance of the final target predictor.

Applied Research

The Australian Artificial Intelligence Institute has engaged in projects with over 60 industry partners, resulting in applied research in various sectors including health care, financial services, business intelligence, logistics, transportation, education, defence and marine safety.

Machine Learning-Based Carriage Load Prediction

Industry partner: Sydney Trains

AAII has successfully applied their concept drift learning model to use Opal card data to predict short-term (2 weeks) and real-time carriage loads.

This intelligent system will support train managers across Sydney’s train network to avoid congestion by directing passengers to the least crowded carriages.

Bionic visual-spatial prosthesis for the blind (CRC-P)

Industry partners: ARIA Research, the University of Sydney, World Access for the Blind Australia Ltd.

Wearable Audio Rendering Devices to assist the visually impaired are limited by human factors (e.g. stress, mental workload, fatigue).

This system incorporates spatial audio and sensory perception, machine/computer vision, visual mapping (visual SLAM) and digital twin and metaverse techniques to render virtual worlds for the visually impaired.

Brain-Robot Interface (BRI) for Defence

Industry partner: Australian Defence Innovation Hub (DIH)

The brain-robot interaction system (BRI) allows for hands-free control of robots - robots can now be controlled entirely by thought. The device is powered by AI and graphene sensors, developed by AAII in collaboration with the Integrated Nano Systems (INSys) Lab.

SharkSpotter: Automatic Shark Detection

Industry Partner: Ripper Corporation

SharkSpotter© is the world's first automatic, real-time AI shark detection system for Unmanned Aerial Vehicles (UAV)/Drones.

SharkSpotter© is more than 90% accurate - current techniques are less than 30% accurate.

AI Based COVID-19 Vaccination Strategy

Industry Partner: Department of Health

AAII played a vital role in identifying priority population cohorts under each phase of the Government’s COVID-19 Vaccination Strategy, using expertise in machine learning and data analytics.

AAII’s involvement informed strategies for achieving Australia’s COVID-19 vaccination rate.

3D Event Reconstruction

Industry Partner: National Institute of Standards and Technology (US)

3D event reconstruction has been made possible with AI’s machine learning and physics models, capabilities AAII has developed over the past few years.

The system synchronises videos from social media and geolocation (e.g. Google street view) to order unstructured videos, resulting in 3D reconstruction.

Facilities

AAII has 8 research labs:

DATA SCIENCES AND KNOWLEDGE DISCOVERY (DSKD Lab)

Machine Learning and Data Mining

The DSKD lab focuses on driving theoretical and practical innovation in data science and knowledge discovery; machine learning; and big data analytics. The Lab develops techniques and tools to help businesses to solve problems and make smarter decisions.

DECISION SYSTEMS AND e-SERVICE INTELLIGENCE (DeSI Lab)

Data-driven decision making through fuzzy logic and machine learning

DeSI Lab aims to develop fundamental knowledge, advanced methodologies and translatable technologies to enable machine learning in complex situations and effectively support data-driven prediction and decision making in real-world systems.

COMPUTATIONAL INTELLIGENCE AND BRAIN-COMPUTER INTERFACE LAB (CIBCI Lab)

Brain-computer interface and EEG assessment systems

CIBCI Centre researchers develop wearable wireless EEG headsets that detect human cognitive stages in real time and provide feedback to improve human performance.

BIOMEDICAL DATA SCIENCE LAB (BDS Lab)

Data analytics in biomedical and clinical domains

The BDS Lab aims to develop computational intelligence for decision making in biomedical and clinical domains. The lab’s research team is helping to predict treatment outcomes for cancer, assisting in vaccine discovery methods for agriculture, producing population-level treatment maps for cancer and developing virtual reality tools to explore oncology data.

LARGE-SCALE NETWORK ANALYTICS LAB (LSN Lab)

Network analytics and solutions for large-scale networks

The Large-Scale Network Analytics Lab focuses on conducting fundamental research to tackle challenging problems in network analytics and to explore future directions in the processing and analysis of large-scale networks.

RECOGNITION, LEARNING AND REASONING LAB (ReLER Lab)

Computer vision and machine intelligence The ReLER Lab is committed to enabling machines to accurately recognise the environment, adaptively understand human interactions and autonomously analyse behaviour through reasoning. Lab researchers are developing novel methods for object, face, action and event recognition.

INTELLIGENT DRONE LAB (iDL)

AI-powered Autonomous drones

The Intelligent Drone Lab (iDL) works with industry to facilitate research and development in drone autonomy using computer vision and machine learning techniques, and a range of sensor inputs.

INTELLIGENT COMPUTING AND SYSTEMS LAB (ICS Lab)

Bioinspired Neural Networks for Deep Learning

The ICS lab studies the characteristics of neuromorphic computing, explores semi-supervised learning data set processing technology and designs and improves deep learning algorithms.

See also

• Artificial Intelligence
• Computer Science
• University of Technology, Sydney

External links

• Official website

This article "Australian Artificial Intelligence Institute" is from Wikipedia. The list of its authors can be seen in its historical and/or the page Edithistory:Australian Artificial Intelligence Institute. Articles copied from Draft Namespace on Wikipedia could be seen on the Draft Namespace of Wikipedia and not main one.

This page exists already on Wikipedia.