Kevin Wong

There are several ways to search for songs in an online music library. A few types of visual variables to represent music information such as colour, position, shape, size, and visual texture have been explored in Music Information Retrieval (MIR). However, from a comprehensive literature review, there is no research focusing explicitly on the use of visual texture for browsing music. In this research, we define visual texture as an image of texture designed using the drawing application. In this paper, a framework for visualising music mood using visual texture is proposed. This proposed framework can be used by designers or software developers to select suitable visual elements when designing a clear and understandable visual texture to represent specific music moods in the music application. This research offers a new way of browsing digital music collection and assisting the music listener community to discover new song especially in mood category. To validate the framework, usability testing was conducted. This paper presents the process of developing and validating the proposed framework.

Digital pathology has attracted significant attention in recent years. Analysis of Whole Slide Images (WSIs) is challenging because they are very large, i.e., of Giga-pixel resolution. Identifying Regions of Interest (ROIs) is the first step for pathologists to analyse further the regions of diagnostic interest for cancer detection and other anomalies. In this paper, we investigate the use of RCNN, which is a deep machine learning technique, for detecting such ROIs only using a small number of labelled WSIs for training. For experimentation, we used real WSIs from a public hospital pathology service in Western Australia. We used 60 WSIs for training the RCNN model and another 12 WSIs for testing. The model was further tested on a new set of unseen WSIs. The results show that RCNN can be effectively used for ROI detection from WSIs.

The performance of classification algorithms with imbalanced streaming data depends upon efficient re-balancing strategy for learning tasks. The difficulty becomes more elevated with multi-class highly imbalanced streaming data. In this paper, we investigate the multi-class imbalance problem in data streams and develop an adaptive framework to cope with imbalanced data scenarios. The proposed One-Vs-All Adaptive Window re-Balancing with Retain Knowledge (OVA-AWBReK) classification framework will combine OVA binarization with Automated Re-balancing Strategy (ARS) using Racing Algorithm (RA). We conducted experiments on highly imbalanced datasets to demonstrate the use of the proposed OVA-AWBReK framework. The results show that OVA-AWBReK framework can enhance the classification performance of the multi-class highly imbalanced data.

Information overloading leads to the need for an efficient search tool to eliminate a considerable amount of irrelevant or unimportant data and present the contents in an easy-browsing form. Personalized faceted search has been one of the potential tools to provide a hierarchical list of facets or categories that helps searchers to organize the information of the search results. Facet selection is one of the important steps to pursue a good faceted search. Collaborative-based personalization was introduced to facet selection. Previous studies have been performed on the use of Collaborative Filtering techniques for personalized facet selection. However, none of the study has investigated Artificial neural network techniques on personalized facet selection. Therefore, this study aims to investigate the possible use of deep Autoencoder on the prediction of facet interests. Autoencoder model was applied to address the association of collaborative interest in facets. The experiments were conducted on 100K and 1M rating records of Movielen dataset. Rating score was used to represent the explicit feedback on facet interests. The performance was reported by comparing the proposed technique and the state-of-the-art model-based Collaborative Filtering techniques in terms of prediction accuracy and computational time. The results showed that the proposed Autoencoder-based model achieved better performance and it was able to significantly improve the prediction of personal facet interests.

In the real world of credit card fraud detection, due to a minority of fraud related transactions, has created a class imbalance problem. With the increase of transactions at massive scale, the imbalanced data is immense and has created a challenging issue on how well Machine Learning (ML) techniques can scale up to efficiently learn to detect fraud from the massive incoming data and to respond faster with high prediction accuracy and reduced misclassification costs. This paper is based on experiments that compared several popular ML techniques and investigated their suitability as a “scalable algorithm” when working with highly imbalanced massive or “Big” datasets. The experiments were conducted on two highly imbalanced datasets using Random Forest, Balanced Bagging Ensemble, and Gaussian Naïve Bayes. We observed that many detection algorithms performed well with medium-sized dataset but struggled to maintain similar predictions when it is massive.

Search engine plays an important role in information retrieval as being the preferred tool by users to locate and manage their desired information. The volume of online data has dramatically increased and this phenomenon of impressive growth leads to the need of efficient systems to deal with issues associated with storage and retrieval. Keyword search is the most popular search paradigm which prompts user to search the entire repository based on a few keywords. From research article collection, using keyword search only may not be enough for researchers to explore academic documents related to their interests from the entire repository. Knowledge discovery tool has recently received much attention in order to compensate the weakness of keyword search usage for academic collection. This paper presents the practical system design and implementation of a knowledge discovery tool in terms of faceted search. This study focused on Thai research articles for use by Thai scholars. The proposed faceted search system was constructed based on the Apache Solr search platform. The methodology of data preparation, knowledge extraction and implementation are discussed in the paper.

Visual design plays an important role in grabbing web users’ attention in an online environment. Previous research has demonstrated that different types of visual design causes different impact towards the end-users. This paper observes the impact of persuasive visual towards users’ first impression, attitudes, and behaviours. It extends existing web visual design by empirically examining the critical roles of the principles of social influence in the form of visual persuasion in motivating users to have a favourable impression of a particular website. Survey data was collected in an experimental study that was conducted online. Structural model assessment is carried out using confirmatory factor analysis (CFA) in conjunction with PLS-SEM analyses. The general analysis of model fit indicates that the two models proposed in this paper surpassed the cut off values for model acceptance with most of the model fit criteria reflects outstanding explanatory power. The result of the analysis indicates that visual persuasion has a big impact in influencing users’ attitudes on the web; strong enough to affect their behavioural intention.

n this paper, a Complementary Fuzzy Support Vector Machine (CMTFSVM) technique is proposed to handle outlier and noise in classification problems. Fuzzy membership values are applied for each input point to reflect the degree of importance of the instances. Datasets from the UCI and KEEL are used for the comparison. In order to confirm the proposed methodology, 40 % random noise is added to the datasets. The experiment results of CMTFSVM are analysed and compared with the Complementary Neural Network (CMTNN). The outcome indicated that the combined CMTFSVM outperformed the CMTNN approach.

Dyslexia is a hidden learning disability that causes difficulties in reading and writing despite average intelligence. Electroencephalogram (EEG) is one of the upcoming methods being researched for identifying unique brain activation patterns in dyslexics. This paper examines pros and cons of existing EEG-based analysis and classification frameworks for dyslexia and recommends optimizations through the findings to assist future research.