Efficient tensor based face recognition

This paper addresses the limitation of current multilinear PCA based techniques, in terms of pro- hibitive computational cost of testing and poor gen- eralisation in some scenarios, when applied to large training databases. We define person-specific eigen-modes to obtain a set of projection bases, wherein a particular basis captures variation across light- ings and viewpoints for a particular person. A new recognition approach is developed utilizing these bases. The proposed approach performs on a par with the existing multilinear approaches, whilst sig- nificantly reducing the complexity order of the testing algorithm.

Supervised subspace learning with multi-class Lagrangian SVM on the Grassmann manifold

Learning robust subspaces to maximize class discrimination is challenging, and most current works consider a weak connection between dimensionality reduction and classifier design. We propose an alternate framework wherein these two steps are combined in a joint formulation to exploit the direct connection between dimensionality reduction and classification. Specifically, we learn an optimal subspace on the Grassmann manifold jointly minimizing the classification error of an SVM classifier. We minimize the regularized empirical risk over both the hypothesis space of functions that underlies this new generalized multi-class Lagrangian SVM and the Grassmann manifold such that a linear projection is to be found. We propose an iterative algorithm to meet the dual goal of optimizing both the classifier and projection. Extensive numerical studies on challenging datasets show robust performance of the proposed scheme over other alternatives in contexts wherein limited training data is used, verifying the advantage of the joint formulation.

Flusser's death rattle

A live film performance using magnificent 16mm featuring Dirk De Bruyn in person. Can an image be sonic and ephemeral in the digital age? Live 3-screen film projection, shadow-play and sound poetry plumbing 35 years of experimental film practice, laying bare those processes of graffiti production splattered across the alleyways and railway lines of the planet’s inner cities but whose performance threatens to become completely hidden inside the computer. Images scratched, dyed, bleached and redrawn by hand are brought together to immerse the audience in an aural-visual rant. Does the analogue answer back to the digital media explosion or merely succumb in an angry death rattle of lost causes? Rev presents a rare opportunity to see one of Australia’s most important experimental filmmakers presenting a unique expanded cinema event.

Untitled : live cinema performance

Over the last 30 years Melbourne-based film-maker, writer and academic Dirk de Bruyn has made numerous experimental, documentary and animation films and videos, continuing to maintain a no-budget, independent, self-funded focus for much of his work. De Bruyns distinctive style entails cut-up collages that draw on animation, found footage and fragments of dialogue - dyeing, painting, incising and stencilling the film strip. Live De Bruyn’s anarchic multi projection performances can involve performance, freeform vocal workouts and De Bruyn, ‘bent over and mouthing into a microphone like a demented seagull, totally involved in the relentlessly unravelling collage of home-processed footage’.Penny Webb. Ian Helliwell provided a live electronic soundtrack.

Projected socioeconomic disparities in the prevalence of obesity among Australian adults

Objective: To project prevalence of normal weight, overweight and obesity by educational attainment, assuming a continuation of the observed individual weight change in the 5-year follow-up of the national population survey, the Australian Diabetes, Obesity and Lifestyle study (AusDiab; 2000–2005).

Methods: Age-specific transition probabilities between BMI categories, estimated using logistic regression, were entered into education-level-specific, incidence-based, multi-state life tables. Assuming a continuation of the weight change observed in AusDiab, these life tables estimate the prevalence of normal weight, overweight and obesity for Australian adults with low (secondary), medium (diploma) and high (degree) levels of education between 2005 and 2025.

Results: The prevalence of obesity among individuals with secondary level educational attainment is estimated to increase from 23% in 2000 to 44% in 2025. Among individuals with a degree qualification or higher, it will increase from 14% to 30%. If all current educational inequalities in weight change could be eliminated, the projected difference in the prevalence of obesity by 2025 between the highest and lowest educated categories would only be reduced by half (to a 6 percentage point difference from 14 percentage points).

Conclusion: We predict that almost half of Australian adults with low educational status will be obese by 2025. Current trends in obesity have the potential to drive an increase in the absolute difference in obesity prevalence between educational categories in future years.

Implications: Unless obesity prevention and management strategies focus specifically on narrowing social inequalities in obesity, inequalities in health are likely to widen.

An Intelligent Human-Computer Interaction System for Decision Support

This paper proposes a novel architecture for
developing decision support systems. Unlike conventional decision support systems, the proposed architecture endeavors to reveal the decision-making process such that humans' subjectivity can be
incorporated into a computerized system and, at the same time, to
preserve the capability of the computerized system in processing information objectively. A number of techniques used in developing the decision support system are elaborated to make the decisionmarking
process transparent. These include procedures for high dimensional data visualization, pattern classification, prediction, and evolutionary computational search. An artificial data set is first
employed to compare the proposed approach with other methods. A simulated handwritten data set and a real data set on liver disease diagnosis are then employed to evaluate the efficacy of the proposed
approach. The results are analyzed and discussed. The potentials of the proposed architecture as a useful decision support system are demonstrated.

Space-for-time substitution to assess likely response of aquatic ecosystems to future climate-related change

Predicting and managing ecological response to a changing climate is often limited by an incomplete understanding of response thresholds and biogeographic differences. For example, step changes in rainfall and runoff, and threshold dynamics and hysteresis in ecological response make projection of future conditions difficult. To combat these constraints we propose that biophysical data across exiting climatic gradients can be used in a space-for-time substitution to predict climate-related ecological response elsewhere. This method builds on previous attempts at space-for-time substitution by using patterns in physical and physicochemical data to explain biological differences across the spatial gradient, then using those patterns to formulate hypotheses of temporal ecological response and finally testing those hypotheses on temporal data available in a second, similar region of interest. 

Visualizing and classifying data using a hybrid intelligent system

In this paper, a hybrid intelligent system that integrates the SOM (Self-Organizing Map) neural network, kMER (kernel-based Maximum Entropy learning Rule), and Probabilistic Neural Network (PNN) for data visualization and classification is proposed. The rationales of this Probabilistic SOM-kMER model are explained, and its applicability is demonstrated using two benchmark data sets. The results are analyzed and compared with those from a number of existing methods. Implication of the proposed hybrid system as a useful and usable data visualization and classification tool is discussed.

NeoGeoNative

NeoGeoNative is about my uneasy relationship with the appropriation of Indigenous motifs in contemporary fashion to which I am both attracted and repelled. Cheap 'triba' leggings that I have purchased from places like Sportsgirl, Glassons and the Preston Market have been edited into a kind of hypnotic/seductive kaleidoscope of colour and patterns - that reference the Melanesian, Polynesian and Native American designs that they appropriate - but at the same time are completely decontextualised, re-emerging as neon mandalas and other sacred geometry.

Dance interrogations

A site-specific solo dance work created and performed by Dianne Reid. The work interacts directly with the audience who must move in and around the dance action in intimate and non-traditional performance locations. The dancer creates the performance in real-time, improvising her movement and text in response to the location and the particular audience members she encounters there. The work also incorporates projected video imagery using the skin as projection screen, and a unique soundscape is created for each performance drawn from the music and effects library of SA composer Stuart Day.

Dance interrogations

A site-specific solo dance work created and performed by Dianne Reid. The work interacts directly with the audience who must move in and around the dance action in intimate and non-traditional performance locations. The dancer creates the performance in real-time, improvising her movement and text in response to the location and the particular audience members she encounters there. The work also incorporates projected video imagery using the skin as projection screen, and a unique soundscape is created for each performance drawn from the music and effects library of SA composer Stuart Day.

Improving Top-N recommendations with user consuming profiles

In this work, we observe that user consuming styles tend to change regularly following some profiles. Therefore, we propose a consuming profile model to capture the user consuming styles, then apply it to improve the Top-N recommendation. The basic idea is to model user consuming styles by constructing a representative subspace. Then, a set of candidate items can be estimated by measuring its reconstruction error from its projection on the representative subspace. The experiment results show that the proposed model can improve the accuracy of Top-N recommendations much better than the state-of-the-art algorithms.

Learning rating patterns for Top-N recommendations

Two rating patterns exist in the user × item rating matrix and influence each other: the personal rating patterns are hidden in each user's entire rating history, while the global rating patterns are hidden in the entire user × item rating matrix. In this paper, a Rating Pattern Subspace is proposed to model both of the rating patterns simultaneously by iteratively refining each other with an EM-like algorithm. Firstly, a low-rank subspace is built up to model the global rating patterns from the whole user × item rating matrix, then, the projection for each user on the subspace is refined individually based on his/her own entire rating history. After that, the refined user projections on the subspace are used to improve the modelling of the global rating patterns. Iteratively, we can obtain a well-trained low-rank Rating Pattern Subspace, which is capable of modelling both the personal and the global rating patterns. Based on this subspace, we propose a RapSVD algorithm to generate Top-N recommendations, and the experiment results show that the proposed method can significantly outperform the other state-of-the-art Top-N recommendation methods in terms of accuracy, especially on long tail item recommendations.