Clustering based multi-label classification for image annotation and retrieval

This paper presents a novel multi-label classification framework for domains with large numbers of labels. Automatic image annotation is such a domain, as the available semantic concepts are typically hundreds. The proposed framework comprises an initial clustering phase that breaks the original training set into several disjoint clusters of data. It then trains a multi-label classifier from the data of each cluster. Given a new test instance, the framework first finds the nearest cluster and then applies the corresponding model. Empirical results using two clustering algorithms, four multi-label classification algorithms and three image annotation data sets suggest that the proposed approach can improve the performance and reduce the training time of standard multi-label classification algorithms, particularly in the case of large number of labels.

Efficient dimensionality reduction and one-class classification for content-based image retrieval

The thesis investigates various machine learning approaches to reducing data dimensionality, and studies the impact of asymmetric data on learning in image retrieval. Efficient algorithms are proposed to reduce the data dimensionality. Integration strategies for one-class classification are designed to address asymmetric data issue and improve retrieval effectiveness.

Empirical study of multi-label classification methods for image annotation and retrieval

This paper presents an empirical study of multi-label classification methods, and gives suggestions for multi-label classification that are effective for automatic image annotation applications. The study shows that triple random ensemble multi-label classification algorithm (TREMLC) outperforms among its counterparts, especially on scene image dataset. Multi-label k-nearest neighbor (ML-kNN) and binary relevance (BR) learning algorithms perform well on Corel image dataset. Based on the overall evaluation results, examples are given to show label prediction performance for the algorithms using selected image examples. This provides an indication of the suitability of different multi-label classification methods for automatic image annotation under different problem settings.

Retrieval

Retrieval was a site-specific performance installation which transformed five floors of the National Library of Australia in Nov-Dec 2010. Devised in collaboration with a team of 30 young performers, the production lead the audience on a quest deep into the library to retrieve priceless cultural memories before all was lost. Louise Morris working in collaboration with co-designer Matthew Aberline and the artistic team created the installed environments for the production. The production won numerous awards including Best New Project- Express Media and Best Original Work- CAT awards.

Retrieval-induced forgetting of performed and observed bizarre and familiar actions

To investigate whether people show retrieval-induced forgetting (RIF) for bizarre and familiar actions that they performed or observed, three experiments were conducted. In Experiment 1, participants performed bizarre and familiar actions with different objects during learning (e.g., pencil: balance the pencil across the cup, sharpen the pencil). They repeatedly performed a set of the bizarre or familiar actions during retrieval practice. After a distracter task, participants’ cued recall was tested. Participants showed RIF for both bizarre and familiar actions. In Experiment 2, half of the participants performed the bizarre and familiar actions themselves; the other half observed the experimenter performing the actions. Replicating the results of Experiment 1, participants who performed the actions showed RIF for bizarre and familiar actions. In contrast, participants who observed the actions did not show RIF for either action type. Experiment 3 examined whether this lack of RIF for observed actions occurred due to a lack of active recall during retrieval practice; it did. Overall, the three experiments demonstrated RIF for both bizarre and familiar performed and observed actions. A distinctiveness account of the results is provided.

Occurrence and distribution of established and new introduced bird species in north Sulawesi, Indonesia

Distributional and habitat information on eight introduced bird species in north Sulawesi, Indonesia, is presented. The accounts are based on our observations as well as being gathered from published sources and unpublished trip reports. Three species (Sulphur-crested Cockatoo Cacatua galerita, Sooty-headed Bulbul Pycnonotus aurigaster and Red-collared Dove Steptopelia tranquebarica) have not previously been reported in north Sulawesi in the published literature, while the continued presence and status of Java Sparrow Padda oryzivora, Zebra Dove Geopelia striata and Rock Dove Columba livia was considered uncertain in the published literature. Further work is required systematically to document the distribution, status and spread of introduced species in the north and other parts of Sulawesi—an imperative from both an economic and conservation perspective.

When more is less : urban remnants support high bird abundance but diversity varies

Urban remnant vegetation, especially where it occurs in public parks, allows for relatively easy access for ongoing biodiversity monitoring. However, relatively little baseline information on bird species distribution and abundance across a range of identifiable urban remnants appears in the published literature. We surveyed the relative abundance and distribution of birds across urban and suburban remnant vegetation in Melbourne, Australia. One hundred and six species were recorded, of which 98 were indigenous. Red wattlebirds had the highest mean relative abundance with 2.94 birds/ha, followed by rainbow lorikeets (2.51), noisy miners (1.93), brown thornbills (1.75) and spotted doves (0.96). There was no obvious trend between overall relative abundance and the size of the remnant, in contrast to species richness which was positively correlated with remnant size. The data revealed that some species were either totally restricted to, or more abundant in, larger remnants and generally absent from smaller remnants. Some of the more common birds (crimson rosella, superb fairy-wren, spotted pardalote and black-faced cuckoo-shrike) recorded during this study were detected at similar densities to those found in comparable vegetation to the east of Melbourne within a largely forested landscape. Other species occurred at much lower densities (e.g., white-browed scrubwren, brown thornbill, eastern yellow robin and grey fantail) or had habitat requirements or ecological characteristics that could place them at risk of further decline or local extinction in the urban area. We identify a suite of bird species of potential conservation concern within Melbourne’s urban landscape. The establishment of repeatable, fixed-point, and long-term monitoring sites will allow for repeat surveying over time and provide an early warning of population declines, or conversely an indication of population increase for other species.

Bird biometrics condition and disease database

Contains biometric measurements (tarsus length, bill length, total head length, body mass) of birds caught at various locations in Australia (though mainly South Australia and Victoria), including data on the viral (mainly Avian Influenza Virus) analysis of cloacal and oropharyngeal swabs, serology of sera samples, and additional data on metabolites and parasite prevalence in swabs, blood cells and sera. Approximately 1500 birds (mainly waterbirds) are sampled annually.

Ecological correlates of the threat of extinction in neotropical bird species

Predicting the threat of extinction aids efficient distribution of conservation resources. This paper utilises a comparative macroecological approach to investigate the threat of extinction in Neotropical birds. Data on ecological variables for 1708 species are analysed using stepwise regression to produce minimum adequate models, first using raw species values and then using independent contrasts (to control for phylogenetic effects). The models differ, suggesting phylogeny has significant effects. The raw species analysis reveals that number of zoogeographical regions occupied, elevational range and utilisation of specialised microhabitats were negatively associated with threat, while minimum elevation and body mass were positively associated, whereas the independent contrasts analysis only identifies zoogeographical regions as important. Confining the analysis to the 582 species restricted to a single zoogeographical region reveals elevational range and number of habitats occupied to be negatively correlated with threat whether the analysis is based on the raw data or on independent contrasts. Analysis of four contrasting zoogeographical regions highlights regional variation in the models. In two Andean regions the threat of extinction declines as the elevation range across which the species occurs increases. In the presence of substantial human populations on high Andean plateaus, a species with a greater elevational range may be more likely to persist at some (relatively) unsettled altitudes. In Central South America, the strongest predictor of threat is minimum elevation of occurrence: species with a lower minimum are less threatened. The minimum elevation result suggests that lowland species experiencing an ecological limit to their minimum elevation (min. elevation >0 m) may be more at risk than those not experiencing such a limit (min. elevation = 0 m). Finally, in southern Amazonia, where there is little altitudinal variation, the only weak predictors of threat are body size, larger species being more threatened, and number of habitats, species occupying more habitats being less threatened. These contrasting results emphasise the importance of undertaking extinction risk analyses at an appropriate geographical scale. Since the models explained only a low percentage of total variance in the data, the effects of human-mediated habitat disturbance across a wide range of habitats may be important.

Geographical variation in bill size across bird species provides evidence for Allen's rule

Allen’s rule proposes that the appendages of endotherms are smaller, relative to body size, in colder climates, in order to reduce heat loss. Empirical support for Allen’s rule is mainly derived from occasional reports of geographical clines in extremity size of individual species. Interspecific evidence is restricted to two studies of leg proportions in seabirds and shorebirds. We used phylogenetic comparative analyses of 214 bird species to examine whether bird bills, significant sites of heat exchange, conform to Allen’s rule. The species comprised eight diverse taxonomic groups—toucans, African barbets, Australian parrots, estrildid finches, Canadian galliforms, penguins, gulls, and terns. Across all species, there were strongly significant relationships between bill length and both latitude and environmental temperature, with species in colder climates having significantly shorter bills. Patterns supporting Allen’s rule in relation to latitudinal or altitudinal distribution held within all groups except the finches. Evidence for a direct association with temperature was found within four groups (parrots, galliforms, penguins, and gulls). Support for Allen’s rule in leg elements was weaker, suggesting that bird bills may be more susceptible to thermoregulatory constraints generally. Our results provide the strongest comparative support yet published for Allen’s rule and demonstrate that thermoregulation has been an important factor in shaping the evolution of bird bills.

Medical image retrieval with query-dependent feature fusion based on one-class SVM

Due to the huge growth of the World Wide Web, medical images are now available in large numbers in online repositories, and there exists the need to retrieval the images through automatically extracting visual information of the medical images, which is commonly known as content-based image retrieval (CBIR). Since each feature extracted from images just characterizes certain aspect of image content, multiple features are necessarily employed to improve the retrieval performance. Meanwhile, experiments demonstrate that a special feature is not equally important for different image queries. Most of existed feature fusion methods for image retrieval only utilize query independent feature fusion or rely on explicit user weighting. In this paper, we present a novel query dependent feature fusion method for medical image retrieval based on one class support vector machine. Having considered that a special feature is not equally important for different image queries, the proposed query dependent feature fusion method can learn different feature fusion models for different image queries only based on multiply image samples provided by the user, and the learned feature fusion models can reflect the different importances of a special feature for different image queries. The experimental results on the IRMA medical image collection demonstrate that the proposed method can improve the retrieval performance effectively and can outperform existed feature fusion methods for image retrieval.

Series feature aggregation for content-based image retrieval

Feature aggregation is a critical technique in content-based image retrieval systems that employ multiple visual features to characterize image content. One problem in feature aggregation is that image similarity in different feature spaces can not be directly comparable with each other. To address this problem, a new feature aggregation approach, series feature aggregation (SFA), is proposed in this paper. In contrast to merging incomparable feature distances in different feature spaces to get aggregated image similarity in the conventional feature aggregation approach, the series feature aggregation directly deal with images in each feature space to avoid comparing different feature distances. SFA is effectively filtering out irrelevant images using individual features in each stage and the remaining images are images that collectively described by all features. Experiments, conducted with IAPR TC-12 benchmark image collection (ImageCLEF2006) that contains over 20,000 photographic images and defined queries, have shown that SFA can outperform the parallel feature aggregation and linear distance combination schemes. Furthermore, SFA is able to retrieve more relevant images in top ranked outputs that brings better user experience in finding more relevant images quickly.

An unified framework based on p-norm for feature aggregation in content- based image retrieval

Feature aggregation is a critical technique in content- based image retrieval systems that employ multiple visual features to characterize image content. In this paper, the p-norm is introduced to feature aggregation that provides a framework to unify various previous feature aggregation schemes such as linear combination, Euclidean distance, Boolean logic and decision fusion schemes in which previous schemes are instances. Some insights of the mechanism of how various aggregation schemes work are discussed through the effects of model parameters in the unified framework. Experiments show that performances vary over feature aggregation schemes that necessitates an unified framework in order to optimize the retrieval performance according to individual queries and user query concept. Revealing experimental results conducted with IAPR TC-12 ImageCLEF2006 benchmark collection that contains over 20,000 photographic images are presented and discussed.