Caribbean copyright showdown

The tiny Caribbean island of Antigua doesn’t make the news very often. Located 1800 kilometres east of Miami, just north of Montserrat, it is best known as a holiday destination for well-heeled Europeans and celebrities with private yachts. Now it is in the headlines for an unexpected reason...

Child-feeding practices of Indian and Australian-Indian mothers

Aims Child-feeding practices may be modifiable risk factors for childhood obesity; however investigation of feeding practices in non-Western populations is scarce. This cross-sectional study examines feeding practices of affluent Indian mothers with children aged 1-5 years residing in Australia and Mumbai, India. The secondary aim was to study the association between maternal and child characteristics and feeding practices. Methods In Australia 230 and in Mumbai 301 mothers completed either a hardcopy or online questionnaire. Self-reported maternal feeding practices (restriction, monitoring, pressure to eat, passive and responsive feeding) were measured using established scales and culturally-specific items. Results Mothers in both samples were equally likely to use non-responsive feeding practices, namely dietary restriction, pressure and passive feeding. Similarly, at least 50% of mothers in both samples did not feed their child responsively (mother decides what and the child decides how much to eat). The only difference observed after controlling for covariates (mothers’ age, BMI, religion, education, questionnaire type, child’s age, birth place, gender, number of siblings, and weight-for-age (WAZ) scores) was that mothers in the Australian sample used higher levels of dietary monitoring (β= 0.2, P= 0.006). Mothers with a higher BMI (OR: 0.84, CI: 0.89-0.99, p=0.03) and following Hinduism (OR: 0.50, CI: 0.33-0.83, p=0.008) were less likely to feed responsively. Conclusions These results suggest that Indian mothers in both the samples may benefit from interventions that promote responsive child-feeding practices.

Determinants and measurement of lean body mass in Indian adults

Indians tend to have lower lean body mass than other ethnic groups which increases the risk of chronic diseases. Three complementary studies included in this thesis advanced knowledge on determinants of lean body mass in Indians and the techniques to measure it. The first study examined the determinants of lean body mass in young Indian adults and highlighted the importance of diet and physical activity for development of lean body mass. This study has important implications for policy on prevention of chronic diseases in India. The other two studies helped refinement of the techniques of lean body mass measurement and are expected to facilitate future research in this area. The thesis is presented in the form of publications in high ranking journals.

Foot morphological difference between habitually shod and unshod runners

Foot morphology and function has received increasing attention from both biomechanics researchers and footwear manufacturers. In this study, 168 habitually unshod runners (90 males whose age, weight & height were 23 +/- 2.4years, 66 +/- 7.1kg & 1.68 +/- 0.13m and 78 females whose age, weight & height were 22 +/- 1.8years, 55 +/- 4.7kg & 1.6 +/- 0.11m) (Indians) and 196 shod runners (130 males whose age, weight & height were 24 +/- 2.6years, 66 +/- 8.2kg & 1.72 +/- 0.18m and 66 females whose age, weight & height were 23 +/- 1.5years, 54 +/- 5.6kg & 1.62 +/- 0.15m)(Chinese) participated in a foot scanning test using the easy-foot-scan (a three-dimensional foot scanning system) to obtain 3D foot surface data and 2D footprint imaging. Foot length, foot width, hallux angle and minimal distance from hallux to second toe were calculated to analyze foot morphological differences. This study found that significant differences exist between groups (shod Chinese and unshod Indians) for foot length (female p = 0.001), width (female p = 0.001), hallux angle (male and female p = 0.001) and the minimal distance (male and female p = 0.001) from hallux to second toe. This study suggests that significant differences in morphology between different ethnicities could be considered for future investigation of locomotion biomechanics characteristics between ethnicities and inform last shape and design so as to reduce injury risks and poor performance from mal-fit shoes.

FreeS: A fast algorithm to discover frequent free subtrees using a novel canonical form

Web data can often be represented in free tree form; however, free tree mining methods seldom exist. In this paper, a computationally fast algorithm FreeS is presented to discover all frequently occurring free subtrees in a database of labelled free trees. FreeS is designed using an optimal canonical form, BOCF that can uniquely represent free trees even during the presence of isomorphism. To avoid enumeration of false positive candidates, it utilises the enumeration approach based on a tree-structure guided scheme. This paper presents lemmas that introduce conditions to conform the generation of free tree candidates during enumeration. Empirical study using both real and synthetic datasets shows that FreeS is scalable and significantly outperforms (i.e. few orders of magnitude faster than) the state-of-the-art frequent free tree mining algorithms, HybridTreeMiner and FreeTreeMiner.

R1STM: One-class support tensor machine with randomised kernel

Identifying unusual or anomalous patterns in an underlying dataset is an important but challenging task in many applications. The focus of the unsupervised anomaly detection literature has mostly been on vectorised data. However, many applications are more naturally described using higher-order tensor representations. Approaches that vectorise tensorial data can destroy the structural information encoded in the high-dimensional space, and lead to the problem of the curse of dimensionality. In this paper we present the first unsupervised tensorial anomaly detection method, along with a randomised version of our method. Our anomaly detection method, the One-class Support Tensor Machine (1STM), is a generalisation of conventional one-class Support Vector Machines to higher-order spaces. 1STM preserves the multiway structure of tensor data, while achieving significant improvement in accuracy and efficiency over conventional vectorised methods. We then leverage the theory of nonlinear random projections to propose the Randomised 1STM (R1STM). Our empirical analysis on several real and synthetic datasets shows that our R1STM algorithm delivers comparable or better accuracy to a state-of-the-art deep learning method and traditional kernelised approaches for anomaly detection, while being approximately 100 times faster in training and testing.