Stationary processes for object detection and non-rigid structure-from-motion

Stationary processes are random variables whose value is a signal and whose distribution is invariant to translation in the domain of the signal. They are intimately connected to convolution, and therefore to the Fourier transform, since the covariance matrix of a stationary process is a Toeplitz matrix, and Toeplitz matrices are the expression of convolution as a linear operator. This thesis utilises this connection in the study of i) efficient training algorithms for object detection and ii) trajectory-based non-rigid structure-from-motion.

Parenting a child with haemophilia while living in a non-metropolitan area

The experience of living in a non-metropolitan area and parenting a child with haemophilia is relatively unknown. Using Interpretive Phenomenological Analysis (IPA), the following study explored the experiences of seven parents, from which four themes emerged: ‘bearing the brunt of diagnosis’ captures the impact of the diagnosis; ‘if you can’t help me, who can?’ reveals experiences with the health system; ‘tackling the challenge of treatment’ encompasses difficulties in adhering to the treatment regime; ‘I need you to understand’ reflects desires for others support and understanding. These themes should be considered when developing support systems and interventions for parents living in non-metropolitan areas.

Two worlds apart: Indigenous community perspectives and Non-Indigenous teacher perspectives on Australian schools

Our evaluation studies of Indigenous school reform begin from a different starting point: listening to, hearing and engaging with the commentaries, voices, narratives and analyses of Indigenous community as they discuss and recount their experiences and current encounters with Australian state schools. Here we undertake a contrastive documentation of the views of Indigenous community members, Elders, parents, education workers, and young people and, indeed, of the views of their non-Indigenous teachers and school principals. This is a dramatic picture of two distinctive cultural lifeworlds, communities and worldviews in contact, of two very different ‘constructions’ by participants of a shared, mutual experience: everyday interaction in the social field of the Australian school. Taken together, our Indigenous and non-Indigenous participants repeatedly confirmed and corroborated a key theme: that Indigenous peoples continue to be viewed and ‘treated’ through the lens and language of cultural, intellectual and moral ‘deficit’.

Composite scheme using localized relaxation with non-standard finite difference method for hyperbolic conservation laws

Non-standard finite difference methods (NSFDM) introduced by Mickens [Non-standard Finite Difference Models of Differential Equations, World Scientific, Singapore, 1994] are interesting alternatives to the traditional finite difference and finite volume methods. When applied to linear hyperbolic conservation laws, these methods reproduce exact solutions. In this paper, the NSFDM is first extended to hyperbolic systems of conservation laws, by a novel utilization of the decoupled equations using characteristic variables. In the second part of this paper, the NSFDM is studied for its efficacy in application to nonlinear scalar hyperbolic conservation laws. The original NSFDMs introduced by Mickens (1994) were not in conservation form, which is an important feature in capturing discontinuities at the right locations. Mickens [Construction and analysis of a non-standard finite difference scheme for the Burgers–Fisher equations, Journal of Sound and Vibration 257 (4) (2002) 791–797] recently introduced a NSFDM in conservative form. This method captures the shock waves exactly, without any numerical dissipation. In this paper, this algorithm is tested for the case of expansion waves with sonic points and is found to generate unphysical expansion shocks. As a remedy to this defect, we use the strategy of composite schemes [R. Liska, B. Wendroff, Composite schemes for conservation laws, SIAM Journal of Numerical Analysis 35 (6) (1998) 2250–2271] in which the accurate NSFDM is used as the basic scheme and localized relaxation NSFDM is used as the supporting scheme which acts like a filter. Relaxation schemes introduced by Jin and Xin [The relaxation schemes for systems of conservation laws in arbitrary space dimensions, Communications in Pure and Applied Mathematics 48 (1995) 235–276] are based on relaxation systems which replace the nonlinear hyperbolic conservation laws by a semi-linear system with a stiff relaxation term. The relaxation parameter (λ) is chosen locally on the three point stencil of grid which makes the proposed method more efficient. This composite scheme overcomes the problem of unphysical expansion shocks and captures the shock waves with an accuracy better than the upwind relaxation scheme, as demonstrated by the test cases, together with comparisons with popular numerical methods like Roe scheme and ENO schemes.

Domains in complex surfaces with non-compact automorphism groups

Let X be an arbitrary complex surface and D a domain in X that has a non-compact group of holomorphic automorphisms. A characterization of those domains D that admit a smooth, weakly pseudoconvex, finite type boundary orbit accumulation point is obtained.

Relationship between quality certification and financial and non-financial performance of organizations

At present, many countries have either embraced ISO9001 or used it as the basis of their national quality certification systems. However, few studies have been conducted to examine the benefits companies’ gain from achieving and implementing ISO9001 standards (Chikuku et al. 2012; Psomas et al. 2013; Sampaio et al. 2011a,b). Analysis has brought much more confused and uneven results across the countries. Turning to the experience of Malaysia, this country has witnessed a spectacular growth at an average rate of 9.89% per annum of ISO certificates issued to companies operating within its borders (ISO Survey 2012). While many companies rush to be ISO 9001 certified whether this brings about better benefits (both the financial and the non-financial) is still an open question. In this study, the research problems were first formulated from the literature and then a questionnaire survey was conducted to test the hypotheses. A survey was administered to chief executives officers and managers across manufacturing and service organizations in Malaysia. Multivariate analysis and SPSS macro developed by Preacher and Hayes were used as statistical techniques to the financial and non-financial benefits of ISO9001 certification. The survey instrument was a two-page questionnaire comprising three sections. The first section of the questionnaire covered the company’s profile. The second section consisted of 25 items on internal benefits and third section consisted of 7 items on external benefits measured on 1–5 Likert scale to assess the benefits of ISO9001 certification. Total 201 valid responses were received. Results of the study indicate that there was no significant direct relationship between ISO9001 certification and organizational financial performance, while strong statistical evidence was found to support the direct relationship between ISO9001 certification and non-financial performance. The findings of the study discovered that financial performance is actually directly related to two non-financial measures, namely quality performance and local and international business performance, which are directly and significantly influenced by ISO9001 certification. Therefore non-financial performance measures are involved in the mediational process. The findings will assist practitioners in taking right courses of action that make the implementation of this standard more effective. For example, the study findings study suggests that companies should put emphasize on nonfinancial factors to improve their financial performance.

Ion pairs in non-redundant protein structures

Ion pairs contribute to several functions including the activity of catalytic triads, fusion of viral membranes, stability in thermophilic proteins and solvent-protein interactions. Furthermore, they have the ability to affect the stability of protein structures and are also a part of the forces that act to hold monomers together. This paper deals with the possible ion pair combinations and networks in 25% and 90% non-redundant protein chains. Different types of ion pairs present in various secondary structural elements are analysed. The ion pairs existing between different subunits of multisubunit protein structures are also computed and the results of various analyses are presented in detail. The protein structures used in the analysis are solved using X-ray crystallography, whose resolution is better than or equal to 1.5 angstrom and R-factor better than or equal to 20%. This study can, therefore, be useful for analyses of many protein functions. It also provides insights into the better understanding of the architecture of protein structure.

Cluster based training for scaling non-linear support vector machines

Support Vector Machines(SVMs) are hyperplane classifiers defined in a kernel induced feature space. The data size dependent training time complexity of SVMs usually prohibits its use in applications involving more than a few thousands of data points. In this paper we propose a novel kernel based incremental data clustering approach and its use for scaling Non-linear Support Vector Machines to handle large data sets. The clustering method introduced can find cluster abstractions of the training data in a kernel induced feature space. These cluster abstractions are then used for selective sampling based training of Support Vector Machines to reduce the training time without compromising the generalization performance. Experiments done with real world datasets show that this approach gives good generalization performance at reasonable computational expense.

Low-delay, high-rate, non-square STBCs from scaled complex orthogonal designs

Space-time block codes (STBCs) obtained from non-square complex orthogonal designs are bandwidth efficient compared to those from square real/complex orthogonal designs for colocated coherent MIMO systems and has other applications in (i) non-coherent MIMO systems with non-differential detection, (ii) Space-Time-Frequency codes for MIMO-OFDM systems and (iii) distributed space-time coding for relay channels. Liang (IEEE Trans. Inform. Theory, 2003) has constructed maximal rate non-square designs for any number of antennas, with rates given by [(a+1)/(2a)] when number of transmit antennas is 2a-1 or 2a. However, these designs have large delays. When large number of antennas are considered this rate is close to 1/2. Tarokh et al (IEEE Trans. Inform. Theory, 1999) have constructed rate 1/2 non-square CODs using the rate-1 real orthogonal designs for any number of antennas, where the decoding delay of these codes is less compared to the codes constructed by Liang for number of transmit antennas more than 5. In this paper, we construct a class of rate-1/2 codes for arbitrary number of antennas where the decoding delay is reduced by 50% when compared with the rate-1/2 codes given by Tarokh et al. It is also shown that even though scaling the variables helps to lower the delay it can not be used to increase the rate.