A5058982639- Topic Modeling
- Web Data Mining and Analysis
- Text and Document Classification Technologies
- Image Retrieval and Classification Techniques
- Information Retrieval and Search Behavior
- Galaxies: Formation, Evolution, Phenomena
- Complex Network Analysis Techniques
- Retinal Imaging and Analysis
- Caching and Content Delivery
- Semantic Web and Ontologies
- Natural Language Processing Techniques
- Recommender Systems and Techniques
- Retinal Diseases and Treatments
- Gamma-ray bursts and supernovae
- Algorithms and Data Compression
- Astrophysics and Cosmic Phenomena
- Data Quality and Management
- Advanced Text Analysis Techniques
- Machine Learning and Data Classification
- Face and Expression Recognition
- Advanced Image and Video Retrieval Techniques
- Advanced Clustering Algorithms Research
- Stochastic processes and financial applications
- Retinal and Optic Conditions
- Advanced Data Compression Techniques
Western Sydney University
2015-2025
The University of Melbourne
2001-2009
ARC Centre of Excellence for Transformative Meta-Optical Systems
2001
Abstract We have found a class of circular radio objects in the Evolutionary Map Universe Pilot Survey, using Australian Square Kilometre Array Pathfinder telescope. The appear images as edge-brightened discs, about one arcmin diameter, that are unlike other previously reported literature. explore several possible mechanisms might cause these objects, but none seems to be compelling explanation.
Objectives: This study aims to develop generic velocity thresholds for the analysis of external load data collected in international women’s football matches.Methods: Doppler-derived recordings instantaneous and acceleration were (10 Hz GPS) from 27 female players during 52 matches between 2012 2015. Data examined with k-means, Gaussian mixture model (GMM), Spectral Clustering methods identify four zones, each completed half match-play (277 observations). was also performed 4 different...
This paper demonstrates a novel and efficient unsupervised clustering method with the combination of self-organizing map (SOM) convolutional autoencoder. The rapidly increasing volume radio-astronomical data has increased demand for machine-learning methods as solutions to classification outlier detection. Major astronomical discoveries are unplanned found in unexpected, making machine learning highly desirable by operating without assumptions labeled training data. Our approach shows SOM...
Retinal arteriovenous (AV) nicking is one of the prominent and significant microvascular abnormalities. It characterized by decrease in venular caliber at both sides an artery-vein crossing. Recent research suggests that retinal AV a strong predictor eye diseases such as branch vein occlusion cardiovascular stroke. In this study, we present novel method for objective quantitative assessment. From input image, vascular network first extracted using multiscale line detection method. The...
Information retrieval systems are evaluated against test collections of topics, documents, and assessments which documents relevant to topics. Documents chosen for relevance assessment by pooling runs from a set existing systems. New can return unassessed leading an evaluation bias them. In this paper, we propose estimate the degree unpooled system, adjust system's score accordingly. Bias estimation be done via leave-one-out experiments on existing, pooled systems, but requires problematic...
Current information retrieval methods either ignore the term positions or deal with exact positions; former can be seen as coarse document resolution, latter fine resolution. We propose a new spectral-based method that is able to utilize many different levels of resolution by examining patterns occur in documents. To do this, we take advantage multiresolution analysis properties wavelet transform. show are achieve higher precision when compared vector space and proximity methods, while...
The concept of recall has been one the key elements system measurement throughout history information retrieval, despite fact that there are many unanswered questions as to its value. In this essay, we review those and explore several further issues affect usefulness recall. particular, ask whether it is reasonable expect be able measure recall; some researchers conflating concepts answer set cardinality; plausible a user would rely on belief "high recall" deeply an list. Combined with...
We present a new optical sample of three Supernova Remnants and 16 Remnant (SNR) candidates in the Large Magellanic Cloud(LMC). These objects were originally selected using deep H$\alpha$, [SII] [OIII] narrow-band imaging. Most newly found are located less dense regions, near or around edges LMC's main body. Together with previously suggested MCSNR J0541-6659, we confirm SNR nature for two additional objects: J0522-6740 MCSNRJ0542-7104. Spectroscopic follow-up observations 12 LMC high...
Current document retrieval methods use a vector space similarity measure to give scores of relevance documents when related specific query. The central problem with these is that they neglect any spatial information within the in question. We present new method, called Fourier Domain Scoring (FDS), which takes advantage this information, via transform, more accurate ordering set. show FDS gives an improvement precision over measures for common case Web like queries, and it similar results...
Latent semantic analysis (LSA) is a generalized vector space method that uses dimension reduction to generate term correlations for use during the information retrieval process. We hypothesized even though establishes between terms, causing degradation in correlation of itself (self-correlation). In this article, we have proven there direct relationship size LSA and self-correlation. also shown by altering self-correlations gain substantial increase precision, while reducing computation required
We propose a new Spectral text retrieval method using the Discrete Cosine Transform (DCT). By taking advantage of properties DCT and by employing fast query compression techniques found in vector space methods (VSM), we show that can process queries as VSM achieve much higher precision.
Web page prefetching techniques are used to address the access latency problem of Internet. To perform successful prefetching, we must be able predict next set pages that will accessed by users. The PageRank algorithm Google is compute popularity a based on their link structure. In this paper, novel PageRank-like proposed for conducting prediction. Two biasing factors adopted personalize PageRank, so it favors more important One factor length time spent visiting and other frequency was...
Retinal vascular landmark points such as branching and crossovers are important features for automatic retinal image matching abnormality detection. These can enable screening of large dataset through the detection network abnormalities (i.e., arteriovenous nicking, vein occlusion) which hypertension cardiovascular disease prediction. Existing methods crossover point use only local information at each pixel without considering to detect positions. This leads misclassification very acute...
Retinal arteriovenous nicking (AV nicking) is the phenomenon where venule compressed or decreases in its caliber at both sides of an crossing. Recent research suggests that retinal AVN associated with hypertension and cardiovascular diseases such as stroke. In this article, we propose a computer method for assessing severity level AV artery-vein (AV) crossing color images. The vascular network first extracted using based on multi-scale line detection. A trimming process then performed to...
The PageRank algorithm is used in Web information retrieval to calculate a single list of popularity scores for each page the Web. These are rank query results when presented user. By using structure entire one score per document, we calculating general score, not particular any community. Therefore, more suited queries. In this paper, introduce form PageRank, multi-resolution community-based scores, where document obtains dependent on given When related specific community, choose associated...
We examine currency options in the double exponential jump-diffusion version of Heston stochastic volatility model for exchange rate. assume, addition, that domestic and foreign interest rates are governed by CIR dynamics. The instantaneous is correlated with dynamics rate return, whereas short-term assumed to be independent its volatility. main result furnishes a semi-analytical formula price European call option hybrid exchange/interest model.