Barry’s research at UNSW (University of New South Wales), during 1998 to 2004, was in the field of artificial intelligence. His PhD was in the formation of low-level visual processing capabilities from initially random retinal arrangements. He also extended nested ripple-down rules to general predicates, placing RDR in a logic programing framework. Some work was also done on the control of a bi-pedal walking robot.
Barry Drake is a successful researcher & experienced technologist. Prior to joining UTS School of Computer Science, Barry was the Senior Principal Engineer and lead researcher for machine learning in Canon Information Systems Research Australia (CISRA). Barry has 13 years of experience conducting research with commercial impact within Canon for several multi-million dollar projects in Machine Learning and general data science and is the inventor on 20 filed patents.
Barry aims “to make a significant positive impact translating Australian research into effective technology in business and government.” Over the years, he has undertaken business case and technology gap analyses for a large range of applications, including: new office workflow solutions, QA for print manufacturing, new features for professional DSLR cameras and video surveillance solutions. He has also produced detailed research project plans to which Canon committed and successfully executed within CISRA under Dr Drake’s supervision.
Can supervise: YES
Barry's research is directed towards algorithms and software for making practical smart systems. He has a focus on fast, efficient, accurate methods. Topics include:
- applications of artificial intelligence and machine learning
- probabilistic inferencing and knowledge compilation
- learning probabilistic models
- fast-efficient near-neighbour searching in ultra-high dimensional spaces (case-based inference)
- complex human-computer interaction.
Barry has particular interests in the domain of health, especially health systems and delivery of health services.
Barry also has experience in other application domains:
- video and image analytics
- document processing
- decision systems
- computer games.
Barry presents guest lectures in computer science and data science. Topics cover hashing, probabilistic graphical models, knowledge compilation and logical satisfiability.
Curiskis, SA, Drake, B, Osborn, TR & Kennedy, PJ 2019, 'An evaluation of document clustering and topic modelling in two online social networks: Twitter and Reddit', Information Processing and Management.View/Download from: Publisher's site
© 2019 Methods for document clustering and topic modelling in online social networks (OSNs) offer a means of categorising, annotating and making sense of large volumes of user generated content. Many techniques have been developed over the years, ranging from text mining and clustering methods to latent topic models and neural embedding approaches. However, many of these methods deliver poor results when applied to OSN data as such text is notoriously short and noisy, and often results are not comparable across studies. In this study we evaluate several techniques for document clustering and topic modelling on three datasets from Twitter and Reddit. We benchmark four different feature representations derived from term-frequency inverse-document-frequency (tf-idf) matrices and word embedding models combined with four clustering methods, and we include a Latent Dirichlet Allocation topic model for comparison. Several different evaluation measures are used in the literature, so we provide a discussion and recommendation for the most appropriate extrinsic measures for this task. We also demonstrate the performance of the methods over data sets with different document lengths. Our results show that clustering techniques applied to neural embedding feature representations delivered the best performance over all data sets using appropriate extrinsic evaluation measures. We also demonstrate a method for interpreting the clusters with a top-words based approach using tf-idf weights combined with embedding distance measures.
Torres Robles, A, Wiecek, E, Cutler, R, Drake, B, Benrimoj, SI, Fernandez-Llimos, F & Garcia Cardenas, M 2019, 'Using dispensing data to evaluate adherence implementation rates in community pharmacy', Frontiers in Pharmacology, vol. 10.View/Download from: UTS OPUS or Publisher's site
Cutler, RL, Torres-Robles, A, Wiecek, E, Drake, B, Van der Linden, N, Benrimoj, SIC & Garcia-Cardenas, V 2019, 'Pharmacist-led medication non-adherence intervention: reducing the economic burden placed on the Australian health care system.', Patient Preference and Adherence, vol. 13, pp. 853-862.View/Download from: UTS OPUS or Publisher's site
Background: Scarcity of prospective medication non-adherence cost measurements for the Australian population with no directly measured estimates makes determining the burden medication non-adherence places on the Australian health care system difficult. This study aims to indirectly estimate the national cost of medication non-adherence in Australia comparing the cost prior to and following a community pharmacy-led intervention. Methods: Retrospective observational study. A de-identified database of dispensing data from 20,335 patients (n=11,257 on rosuvastatin, n=6,797 on irbesartan and n=2,281 on desvenlafaxine) was analyzed and average adherence rate determined through calculation of PDC. Included patients received a pharmacist-led medication adherence intervention and had twelve months dispensing records; six months before and six months after the intervention. The national cost estimate of medication non-adherence in hypertension, dyslipidemia and depression pre- and post-intervention was determined through utilization of disease prevalence and comorbidity, non-adherence rates and per patient disease-specific adherence-related costs. Results: The total national cost of medication non-adherence across three prevalent conditions, hypertension, dyslipidemia and depression was $10.4 billion equating to $517 per adult. Following enrollment in the pharmacist-led intervention medication non-adherence costs per adult decreased $95 saving the Australian health care system and patients $1.9 billion annually. Conclusion: In the absence of a directly measured national cost of medication non-adherence, this estimate demonstrates that pharmacists are ideally placed to improve patient adherence and reduce financial burden placed on the health care system due to non-adherence. Funding of medication adherence programs should be considered by policy and decision makers to ease the current burden and improve patient health outcomes moving forward.
Montgomery, J, Reid, M & Drake, BJ 2015, 'Protocols and Structures for Inference: A RESTful API for Machine Learning', Proceedings of The 2nd International Conference on Predictive APIs and Apps, 2nd International Conference on Predictive APIs and Apps, Journal of Machine Learning Research, Sydney, pp. 29-42.View/Download from: UTS OPUS
Diversity in machine learning APIs (in both software toolkits and web services), works against realising machine learning's full potential, making it difficult to draw on individual algorithms from different products or to compose multiple algorithms to solve complex tasks. This paper introduces the Protocols and Structures for Inference (PSI) service architecture and specification, which presents inferential entities—relations, attributes, learners and predictors—as RESTful web resources that are accessible via a common but flexible and extensible interface. Resources describe the data they ingest or emit using a variant of the JSON schema language, and the API has mechanisms to support non-JSON data and future extension of service features.
Peters, MW & Drake, BJ 2003, 'The Representation of Space and the Space of Representation: a Cognitive Science Introduction to JIGSAW', Joint International Conference on Cognitive Science, Sydney, pp. 519-524.
Peters, MW & Drake, BJ 2000, 'Fundamental Robust Self-Organising Vision', Proceedings of the Fourth Japan-Australia Joint Workshop on Intelligent and Evolutionary systems, 4th Japan-Australia Joint Workshop on Intelligent and Evolutionary Systems, Yokosuka, Japan.
Mann, G, Armstrong, B, Preston, P & Drake, BJ University of New South Wales 2001, A Dynamically-Balanced Walking Biped, no. UNSW-CSE-TR-0110.
Peters, MW & Drake, BJ School of Computer Science and Engineering, University of New South Wales. 2000, Jigsaw: the unsupervised construction of spatial representations, no. UNSW-CSE-TR-0007.
Drake, BJ 2004, 'Intelligence, Redundancy and Space'.
The importance of spatial information is manifested by the amount of effort spent collecting and maintaining it. Spatial representations are ubiquitous in both natural and artificial systems although the spatial nature of data is often subtle and neglected. A problem exists in that spatial information is sometimes unknown, ambiguous or just
The question asked by this research is, "Given a set of signals, how can the spatial arrangement of the sources be automatically inferred, without knowledge of what the signal values are supposed to be?" Pursuing this question has led to the investigation of the phenomenon of spatial redundancy and the development of a computer algorithm, JIGSAW.
JIGSAW is a novel method for the recovery of spatial information from a set of signals. It is similar in spirit to Self Organising Maps, but uses ideas from Multidimensional Scaling and Variograms. The developed algorithm embodies efficient methods for the incremental, adaptive construction of a spatial representation for signals. It is scalable and can deal with vast amounts of input data.
JIGSAW has been extensively applied to the domain of computer vision where the task is to discover the relative locations of pixels in digital video. A particular application within this domain is the registration or merging of images from multiple video cameras. It has also been applied to the discovery of the relative locations of weather stations based on meteorological variables, and to the calibration of a robotic pan-tilt device to an image space defined by a camera.
JIGSAW is designed as a general algorithm for any set of signals or data that exhibit redundancy. This work discusses the extendibility of JIGSAW to other sensor types, and to the analysis and presentation of vast amounts of data.