Intelligent Information Technologies for Innovation, Interaction and Creativity in Complex Data Modelling and Decision Support
The programme will develop novel information methods and will apply them for innovative IIT for complex data modelling and decision support systems across application areas. The programme is organised in three tiers: Generic Science Methods; Research Themes; Specific Technology Innovation Projects. In the core of the programme is the fundamental (generic) science of computational intelligence, including, but not limited to: neurocomputation; evolutionary computation; machine learning; data mining; data analytics; decision support; scientific data visualisation; parallel and distributed computing.
Neurocomputing for machine learning and predictive modelling on spatio/spectro temporal data.
The research here will continue the development of a novel neuromorphic information processing architecture called NeuCube and related methodologies for machine learning from spatio-temporal data and for early event prediction. It will be led by N.Kasabov and will involve all NZ participants, in collaboration with the Manchester SpiNNaker team led by S.Furber, the China Academy of Sciences team led by Z.Hou, and other overseas participants - Indiveri, Cihotzki, Ozawa, Yang. This theme will provide methods and platforms to support most of the specific projects in the INTELLECTE programme as explained further in this section.
Stream- and large data modelling and mining.
We will explore a new methodology for stream data analysis based on neurocomputing and drift analysis. The team includes Pears, Kasabov, Gulyaev, Pfahringer, Dobbie. This methodology will be used in several specific projects.
Computer vision and image technologies.
We will explore new methods for fast video and image scene data analysis, including neuromorphic computation with dynamic vision sensors (DVS), also called artificial retinas. The theme will include W.Yan, R.Klette, D.Bailey, N.Kasabov.
Intelligent Human-Computer Interaction.
Our team includes C.Walker, F.Joseph, M.Billinghurst, M.Sagar, R.Jones, S.Marks, S.Weddell and will explore new techniques for visualisation, augmented and virtual reality, wearable computers, affective computing.
Smart systems design and optimization.
Our team will explore a new collaborative approach to distributed software/hardware system design. The team includes T.Clear, S.MacDonell, S.Salcic, M.Winikoff, A.Connor, S.Singamneni, A.Malik. We will also explore methods for complex system optimization at the design phase, such as 3D printing design, including meta evolutionary computation and quantum-inspired computation.
The methods above will be applied for the development of 12 highly interactive and highly collaborative Specific Technology Innovation Projects. Each project will be conducted by a collaborative team including established researchers, emerging researchers and postgraduate students. Each project is a feasibility study on the creation of an innovative IIT to solve a specific and well defined problem with a potential to be further extended. Each project will deliver: (1) a pilot prototype system; (2) a report on the feasibility analysis: (3) at least one journal paper submitted.
1. Novel brain-computer interfaces (BCI).
We will test the hypothesis that a NeuCube-based methodology is more accurate than existing methods for EEG- or fMRI based BCI. This will be tested on neurorehabilitation and brain state recognition tasks. The team includes D.Taylor, N.Kasabov, Shepherd, E.Gaeta, R.Jones, S.Weddel, I.Khan and PhD students Capecci, Scott, Sengupta, Gholami, Chamberlain. Initial experiments have already shown good results. Data will be collected in the EEG lab of the FHES. Successful models will be implemented after 2014 on the robotic system of Z.Hou as part of the China Strategic Alliance Agreement, with a commercialisation prospect.
2. Novel motion data analysis technology.
The first part of the project will test if a NeuCube-based approach will be suitable for modelling and understanding of human motion data. Here the team includes Taylor, Kasabov, McNair, Naude, Signal, PhD students Capecci and Scott. Data will be collected in the GATE lab of the FHES. If successful, results can be expected to be used after 2014 in clinical practice and for sport performance evaluation at AUT. A second part of the project will evaluate the suitability of the NeuCube approach for data-intensive and long–term patient monitoring with the involvement of Lowe, Hosseini, Parry, Harrison (U Otago) and students Baig and Hamano.
3. Predictive personalised modelling of non-communicable diseases.
This project will apply the provisional patent on personalised modelling developed at KEDRI, on temporal or spatio-temporal data and will include case studies on stroke, diabetes and CVD. In its first part, this project will further prove the hypothesis that a NeuCube-based model is more accurate than existing methods for early prediction of stroke occurrence. The team on stroke experiments includes V.Feigin, R.Krishnamurti, N.Kasabov, Z.Hou, postdoc, PhD students M.Othman and V. Breen. Test data is already available for NZ and new data for China will be trialled next year. The project will also explore personalised modelling for risk prediction of diabetes and its CVD implications. This part will involve A.Lowe, Al-Jumaily, Murphy and Plank (U Auckland), Orr and Beban (ADHB).
4. Predicting response to treatment.
This project will test the hypothesis that a NeuCube-based methodology is more accurate than existing methods for the prediction of response to treatment in the health sector. It will be tested on predicting the response to Methadone of people with addiction based on EEG data. The team includes G. Wang, N.Kasabov, R.Kydd, B.Russel, a postdoc and PhD students E.Capecci and M. Gholami. Test data has already been collected.
5. Personalised modelling in Bioinformatics.
This project will explore the KEDRI patented method for personalised modelling on large and heterogeneous data sets that include both genetic and clinical information. The team, that includes C.Higgins, N.Kasabov, D.Love, D.Parry, and PhD V.Breen, will develop a generic tool and explore it on case study problems related to cancer and CVD diagnosis and prognosis. Data will be provided by Auckland Hospital (D. Love) and PEBL (U Otago).
6. Predictive modelling on ecological and environmental data.
This project will further test the hypothesis that a NeuCube-based methodology is more accurate than existing methods for the prediction of the establishment of ecological and environmental events. The first case study will be predicting the establishment of harmful species in NZ. NZ data has already been initially experimented and data from China will be trialled next as part of a tripartite project West China-East China–New Zealand. The second case study will be on the problem of early prediction of risk of earthquakes in NZ based on continuous stream of data collected through the GeoNet system. The team includes Kasabov, Pears, Worner, Yang, Ozawa, a postdoc and PhD students Hartono and Kristiani.
7. A new approach to pattern recognition in radio-astronomy.
This project will develop and test a new method based on the NeuCube architecture for pattern recognition of Pulsars from fast and large streams of radio-astronomy data. If successful, a model implementation on a highly parallel neuromorphic hardware SpiNNaker is envisaged and a proposal to the SKA multibillion programme. The team includes S.Gulyaev, N.Kasabov, R.Pears, B.Pfahringa, G.Dobbie, postdoc and PhD students Scott, Hartono and Kristiani.
8. Computer vision and image processing for dynamic data analysis.
This project will test a methodology of using new methods including dynamic vision sensors (DVS) in combination with a neuromorphic classification system for the purpose of detecting fast moving objects and for face recognition based on video data. The project includes W.Yan, R.Klette, D.Bailey, N.Kasabov, G.Indiveri and a PhD student Fahad Alvi.
9. Visualisation of scientific data.
The project will be a joint effort of CoLab, Sentience Lab, and KEDRI in the development of a generic VR system for the visualisation and understanding of the 3D NeuCube connectivity and dynamics during and after training with spatio-temporal data. First steps have been already done on EEG brain data, but the system will be extended to be used when other types of complex data is being modelled (e.g. human motion; ecological and environmental; medical). The team includes Marks, Estevez, Kasabov, Taylor and student Hartono.
10. Novel human-computer interfaces.
This project will explore the applicability of affective computing based on M.Sagar’s emotional baby model and of new wearable computing devices, such as the Goggle Glasses by M.Billinghurst, in the design of novel IIT. The project will be conducted by C.Walker, F.Joseph, M.Sagar, M.Billinghurst, N.Kasabov, a PhD co-supervised by CoLab/KEDRI.
11. Complex system optimisation.
This project will explore several methods for complex system optimisation, including: evolutionary computation; quantum inspired optimisation; new types of smart meta-heuristic search algorithms. The methods will be tested on the problem of 3D printing design optimisation. The project will involve A.Connor, S.Singamneni, A.Roychoudhry (India), M.Zhang, N.Kasabov and PhD students F.Schmidt and B.Huang.
12. Collaborative and distributed systems design.
This project will integrate the work of all Themes and projects in a new, experimental IIT design approach. A framework will be developed for a collaborative work of several remotely located groups in the design of a distributed software/hardware system that will include elements of: data collection from many distributed sources including video and image data; machine learning of a neurocomputing classifier; system parameter optimisation; human-computer interaction; data flow and system operation visualisation. A case study specification will be decided during the INTELLECTE workshop in May. The team includes T.Clear, S.MacDonell, Z.Salcic, A.Malik, A.Connor, A/Prof.D.Damian (Canada), Da Zhang (PhD student) and the rest of the INTELLECTE team.