An information-theoretic approach to face recognition from face motion manifolds

In this work, we consider face recognition from face motion manifolds (FMMs). The use of the resistor-average distance (RAD) as a dissimilarity measure between densities confined to FMMs is motivated in the proposed information-theoretic approach to modelling face appearance. We introduce a kernel-based algorithm that makes use of the simplicity of the closed-form expression for RAD between two Gaussian densities, while allowing for modelling of complex and nonlinear, but intrinsically low-dimensional manifolds. Additionally, it is shown how geodesically local FMM structure can be modelled, naturally leading to a stochastic algorithm for generalizing to unseen modes of data variation. Recognition performance of our method is demonstrated experimentally and is shown to exceed that of state-of-the-art algorithms. Recognition rate of 98% was achieved on a database of 100 people under varying illumination

Birds in agricultural mosaics : the influence of landscape pattern and countryside heterogeneity

Agricultural environments are critical to the conservation of biota throughout the world. Efforts to identify key influences on the conservation status of fauna in such environments have taken complementary approaches. Many studies have focused on the role of remnant or seminatural vegetation and emphasized the influence on biota of spatial patterns in the landscape. Others have recognized that many species use diverse ‘‘countryside’’ elements within farmland, and emphasize the benefits of landscape heterogeneity for conservation. Here, we investigated the effect of independent measures of both the spatial pattern (extent and configuration) and heterogeneity of elements (i.e., land uses/vegetation types) on bird occurrence in farm-scale agricultural mosaics in southeastern Australia. Birds were sampled in all types of elements in 27 mosaics (each 1 3 1 km) selected to incorporate variation in cover of native vegetation and the number of different element types in the mosaic. We used an information-theoretic approach to identify the mosaic properties that most strongly influenced bird species richness. Subgroups of birds based on habitat requirements responded most strongly to the extent of preferred elements in mosaics. Woodland birds were richer in mosaics with higher cover of native vegetation while open-tolerant species responded to the extent of scattered trees. In contrast, for total species richness, mosaic heterogeneity (richness of element types) and landscape context (cover of native vegetation in surrounding area) had the greatest influence. These results showed that up to 76% of landscape-level variation in richness of bird groups is attributable to mosaic properties directly amenable to management by landowners. Key implications include (1) conservation goals for farm landscapes must be carefully defined because the richness of different faunal components is influenced by different mosaic properties; (2) the extent of native vegetation is a critical influence in agricultural environments because it drives the farmscale richness of woodland birds and has a broader context effect on total bird richness in mosaics; (3) land-use practices that enhance the heterogeneity of farmland mosaics are beneficial for native birds; and (4) the cumulative effect of even small elements in farm mosaics contribute to the structural properties of entire landscapes.

Lecturing and other face-to-face teaching – too much or too little? An assessment based on student feedback and fail rates

Institutes of higher learning are tending to reduce the amount of face-to-face teaching that they offer, and particularly through the traditional pedagogical method of lecturing. There is ongoing debate about the educational value of lectures as a teaching approach, in terms of both whether they facilitate understanding of subject material and whether they augment the student educational experience. In this study, student evaluation of teaching scores plus academic outcome (percentage of students who fail) was assessed for 236 course units offered by a science faculty at an Australian university over the course of one year. These measures were related to the degree to which lectures and other face-to-face teaching were used in these units, controlling for factors such as class size, school and year level. An information-theoretic model selection approach was employed to identify the best models and predictors of student assessments and fail rates. All the top models of student feedback included a measure reflecting amount of face-to-face teaching, with the evaluation of quality of teaching being higher in units with higher proportions of lectures. However, these models explained only 12–20% of the variation in student evaluation scores, suggesting that many other factors come into play. By contrast, units with fewer lectures have lower failure rates. These results suggest that moving away from lectures and face-to-face teaching may not harm, and indeed may improve the number of students who pass the subject, but that this may be incurred at the expense of greater dissatisfaction in students' learning experience.

Spatial prediction of rufous bristlebird habitat in a coastal heathland: a GIS-based approach

1. To develop a conservation management plan for a species, knowledge of its distribution and spatial arrangement of preferred habitat is essential. This is a difficult task, especially when the species of concern is in low   abundance. In south-western Victoria, Australia, populations of the rare rufous bristlebird Dasyornis broadbenti are threatened by fragmentation of suitable habitat. In order to improve the conservation status of this species, critical habitat requirements must be identified and a system of corridors must be established to link known populations. A predictive spatial model of rufous bristlebird habitat was developed in order to identify critical areas requiring preservation, such as corridors for dispersal.
2. Habitat models generated using generalized linear modelling techniques can assist in delineating the specific habitat requirements of a species. Coupled with geographic information system (GIS) technology, these models can be extrapolated to produce maps displaying the spatial configuration of suitable habitat.
3. Models were generated using logistic regression, with bristlebird presence or absence as the dependent variable and landscape variables, extracted from both GIS data layers and multispectral digital imagery, as the predictors. A multimodel inference approach based on Akaike’s information criterion was used and the resulting model was applied in a GIS to extrapolate predicted likelihood of occurrence across the entire area of concern. The predictive performance of the selected model was evaluated using the receiver operating characteristic (ROC) technique. A hierarchical partitioning protocol was used to identify the predictor variables most likely to influence variation in the dependent variable. Probability of species presence was used as an index of habitat suitability.
4. Negative associations between rufous bristlebird presence and  increasing elevation, 'distance to cree', 'distance to coast' and sun index were evident, suggesting a preference for areas relatively low in altitude, in close proximity to the coastal fringe and drainage lines, and receiving less direct sunlight. A positive association with increasing habitat complexity also suggested that this species prefers areas containing high vertical density of vegetation.
5. The predictive performance of the selected model was shown to be high (area under the curve 0·97), indicating a good fit of the model to the data. Hierarchical partitioning analysis showed that all the variables considered had significant  independent contributions towards explaining the variation in the dependent variable. The proportion of the total study area that was predicted as suitable habitat for the rufous bristlebird (using probability of occurrence at a ≥0·5 level ) was 16%.
6. Synthesis and applications. The spatial model clearly delineated areas predicted as highly suitable rufous bristlebird habitat, with evidence of potential corridors linking coastal and inland populations via gullies. Conservation of this species will depend on management actions that protect the critical habitats identified in the model. A multi-scale  approach to the modelling process is recommended whereby a spatially explicit model is first generated using landscape variables extracted from a GIS, and a second model at site level is developed using fine-scale habitat variables measured on the ground. Where there are constraints on the time and cost involved in measuring finer scale variables, the first step alone can be used for conservation planning.

Landscape characteristics associated with species richness and occurrence of small native mammals inhabiting a coastal heathland: a spatial modelling approach

In the coastal region of south-western Victoria, Australia, populations of native small mammal species are restricted to patches of suitable habitat in a highly fragmented landscape. The size and spatial arrangement of these patches is likely to influence both the occupancy and richness of species at a location. Geographic Information System (GIS)-based habitat models of the species richness of native small mammals, and individual species  occurrences, were developed to produce maps displaying the spatial  configuration of suitable habitat. Models were generated using either generalised linear Poisson regression (for species richness) or logistic regression (for species occurrences) with species richness or  presence/absence as the dependent variable and landscape variables, extracted from both GIS data layers and multi-spectral digital imagery, as the predictor variables. A multi-model inference approach based on the Akaike Information Criterion was used and the resulting model was applied in a GIS framework to extrapolate predicted richness/likelihood of occurrence across the entire area of the study. A negative association between species  richness and elevation, habitat complexity and sun index indicated that richness within the study area decreases with increasing altitude, vertical vegetation structure and exposure to solar radiation. Landform  characteristics were important (to varying degrees) in determining habitat occupancy for all of the species examined, while the influence of habitat complexity was important for only one of the species. Performance of all but one of the models generated using presence/absence data was high, as indicated by the area under the curve of a receiver-operating characteristic plot. The effective conservation of the small mammal species in the area of concern is likely to depend on management actions that promote the protection of the critical habitats identified in the models.

An improved approach for the discovery of causal models via MML

Discovering a precise causal structure accurately reflecting the given data is one of the most essential tasks in the area of data mining and machine learning. One of the successful causal discovery approaches is the information-theoretic approach using the Minimum Message Length Principle[19]. This paper presents an improved and further experimental results of the MML discovery algorithm. We introduced a new encoding scheme for measuring the cost of describing the causal structure. Stiring function is also applied to further simplify the computational complexity and thus works more efficiently. The experimental results of the current version of the discovery system show that: (1) the current version is capable of discovering what discovered by previous system; (2) current system is capable of discovering more complicated causal models with large number of variables; (3) the new version works more efficiently compared with the previous version in terms of time complexity.

Occurrence of small mammals in a fragmented landscape: the role of vegetation heterogeneity

Most studies of habitat use by small mammals rely on data from trapping grids. Such studies pertain to habitat use by individuals, which may not reflect population-level requirements. To meet the challenge of landscape change, it is important to understand habitat use by populations across large geographic areas. We surveyed small mammals in 48 forest remnants across a 300 km2 study area, to investigate the influence of vegetation heterogeneity on regional distributions. Information-theoretic techniques were used to evaluate models of vegetation associations. Richness of native mammals was influenced by vegetation condition: disturbed sites supported fewer species. Models for individual species showed the agile antechinus, Antechinus agilis, to prefer structurally diverse forest vegetation, the long-nosed potoroo, Potorous tridactylus, to favour mesic shrub communities, the bush rat, Rattus fuscipes, to prefer complex low cover regardless of composition, the swamp rat, Rattus lutreolus, to favour reduced canopy cover, and the house mouse, Mus domesticus, to prefer disturbed vegetation. To satisfy the needs of all native species, a mosaic of natural vegetation is required. Degradation and simplification of forest vegetation have detrimental consequences. These results highlight the need to consider habitat quality, together with more traditional biogeographic variables, when investigating factors influencing patch occupancy by native fauna in modified landscapes.

Habitat requirements of the yellow-footed antechinus (Antechinus flavipes) in box–ironbark forest, Victoria, Australia

Understanding the habitat requirements of a species is critical for effective conservation-based management. In this study, we investigated the influence of forest structure on the distribution of the yellow-footed antechinus (Antechinus flavipes), a small dasyurid marsupial characteristic of dry forests on the inland side of the Great Dividing Range, Australia. Hair-sampling tubes were used to determine the occurrence of A. flavipes at 60 sites stratified across one of the largest remaining tracts of dry box–ironbark forest in south-eastern Australia. We considered the role of six potential explanatory variables: large trees, hollow-bearing trees, coppice hollows, logs, rock cover and litter. Logistic regression models were examined using an information-theoretic approach to determine the variables that best explained the presence or absence of the species. Hierarchical partitioning was employed to further explore relationships between occurrence of A. flavipes and explanatory variables. Forest structure accounted for a substantial proportion of the variation in occurrence of A. flavipes between sites. The strongest influence on the presence of A. flavipes was the cover of litter at survey sites. The density of hollow-bearing trees and rock cover were also positive influences. The conservation of A. flavipes will be enhanced by retention of habitat components that ensure a structurally complex environment in box–ironbark forests. This will also benefit the conservation of several threatened species in this dry forest ecosystem.

Changing philosophies and tools for statistical inferences in behavioral ecology

Recent developments in ecological statistics have reached behavioral ecology, and an increasing number of studies now apply analytical tools that incorporate alternatives to the conventional null hypothesis testing based on significance levels. However, these approaches continue to receive mixed support in our field. Because our statistical choices can influence research design and the interpretation of data, there is a compelling case for reaching consensus on statistical philosophy and practice. Here, we provide a brief overview of the recently proposed approaches and open an online forum for future discussion (https://bestat.ecoinformatics.org/). From the perspective of practicing behavioral ecologists relying on either correlative or experimental data, we review the most relevant features of information theoretic approaches, Bayesian inference, and effect size statistics. We also discuss concerns about data quality, missing data, and repeatability. We emphasize the necessity of moving away from a heavy reliance on statistical significance while focusing attention on biological relevance and effect sizes, with the recognition that uncertainty is an inherent feature of biological data. Furthermore, we point to the importance of integrating previous knowledge in the current analysis, for which novel approaches offer a variety of tools. We note, however, that the drawbacks and benefits of these approaches have yet to be carefully examined in association with behavioral data. Therefore, we encourage a philosophical change in the interpretation of statistical outcomes, whereas we still retain a pluralistic perspective for making objective statistical choices given the uncertainties around different approaches in behavioral ecology. We provide recommendations on how these concepts could be made apparent in the presentation of statistical outputs in scientific papers.

Landscape-scale effects of fire on bird assemblages : does pyrodiversity beget biodiversity?

Aim A common strategy for conserving biodiversity in fire-prone environments is to maintain a diversity of post-fire age classes at the landscape scale, under the assumption that 'pyrodiversity begets biodiversity'. Another strategy is to maintain extensive areas of a particular seral state regarded as vital for the persistence of threatened species, under the assumption that this will also cater for the habitat needs of other species. We investigated the likely effects of these strategies on bird assemblages in tree mallee vegetation, characterized by multi-stemmed Eucalyptus species, where both strategies are currently employed.

Location The semi-arid Murray Mallee region of south-eastern Australia.

Methods We systematically surveyed birds in 26 landscapes (each 4-km diameter), selected to represent gradients in the diversity of fire age classes and the proportion of older vegetation (>35years since fire). Additional variables were measured to represent underlying vegetation- or fire-mediated properties of the landscape, as well as its biogeographic context. We used an information-theoretic approach to investigate the relationships between these predictor variables and the species richness of birds (total species, threatened species and rare species).

Results Species richness of birds was not strongly associated with fire-mediated heterogeneity. Species richness was associated with increasing amounts of older vegetation in landscapes, but not with the proportion of recently burned vegetation in landscapes.

Main conclusions The preference of many mallee birds for older vegetation highlights the risk of a blanket application of the 'pyrodiversity begets biodiversity' paradigm. If application of this paradigm involved converting large areas from long unburned to recently burned vegetation to increase fire-mediated heterogeneity in tree mallee landscapes, our findings suggest that this could threaten birds. This research highlights the value of adopting a landscape-scale perspective when evaluating the utility of fire-management strategies intended to benefit biodiversity.

Adaptive depletion for improvement of MPEG video compression

Traditional data compression algorithms for 2D images work using the information theoretic paradigm, attempting to reduce redundant information by as much as possible. However, through the use of a depletion algorithm that takes advantage of characteristics of the human visual system, images can be displayed using only half or a quarter of the original information with no appreciable loss of quality.

The characteristics of the human visual system that allows the viewer to perceive a higher rate of information than is actually displayed is known as the beta or picket fence effect. It is called the picket fence effect because its effect is noticeable when a person is travelling along a picket fence. Despite the person not having an unimpeded view of the objects behind the fence at any instant, as the person is moving, the objects behind the picket fence are clearly visible. In fact, in most cases the fence is hardly noticeable at all.

The techniques we have developed uses this effect to achieve higher levels of compression than would otherwise be possible. As a fundamental characteristic of the beta effect is the requirement that there is movement of the fence in relation to the object, the beta effect can only be used in image sequences where movement between the depletion pattern and objects within the image can be achieved.

As MPEG is the recognised standard by which image sequences are coded, compatibility with MPEG is essential. We have modified our technique such that it performs in conjunction with MPEG, providing further compression over MPEG.

Learning Boltzmann distance metric for face recognition

We introduce a new method for face recognition using a versatile probabilistic model known as Restricted Boltzmann Machine (RBM). In particular, we propose to regularise the standard data likelihood learning with an information-theoretic distance metric defined on intra-personal images. This results in an effective face representation which captures the regularities in the face space and minimises the intra-personal variations. In addition, our method allows easy incorporation of multiple feature sets with controllable level of sparsity. Our experiments on a high variation dataset show that the proposed method is competitive against other metric learning rivals. We also investigated the RBM method under a variety of settings, including fusing facial parts and utilising localised feature detectors under varying resolutions. In particular, the accuracy is boosted from 71.8% with the standard whole-face pixels to 99.2% with combination of facial parts, localised feature extractors and appropriate resolutions.

Learning sparse latent representation and distance metric for image retrieval

The performance of image retrieval depends critically on the semantic representation and the distance function used to estimate the similarity of two images. A good representation should integrate multiple visual and textual (e.g., tag) features and offer a step closer to the true semantics of interest (e.g., concepts). As the distance function operates on the representation, they are interdependent, and thus should be addressed at the same time. We propose a probabilistic solution to learn both the representation from multiple feature types and modalities and the distance metric from data. The learning is regularised so that the learned representation and information-theoretic metric will (i) preserve the regularities of the visual/textual spaces, (ii) enhance structured sparsity, (iii) encourage small intra-concept distances, and (iv) keep inter-concept images separated. We demonstrate the capacity of our method on the NUS-WIDE data. For the well-studied 13 animal subset, our method outperforms state-of-the-art rivals. On the subset of single-concept images, we gain 79:5% improvement over the standard nearest neighbours approach on the MAP score, and 45.7% on the NDCG.