scispace - formally typeset
Search or ask a question
Institution

Macquarie University

EducationSydney, New South Wales, Australia
About: Macquarie University is a education organization based out in Sydney, New South Wales, Australia. It is known for research contribution in the topics: Population & Context (language use). The organization has 14075 authors who have published 47673 publications receiving 1416184 citations. The organization is also known as: Macquarie uni.


Papers
More filters
Book ChapterDOI
08 Oct 2016
TL;DR: This paper proposed a new automated caption evaluation metric defined over scene graphs coined SPICE, which captures human judgments over model-generated captions better than other automatic metrics (e.g., system-level correlation of 0.88 with human judgments on the MS COCO dataset, versus 0.43 for CIDEr and 0.53 for METEOR).
Abstract: There is considerable interest in the task of automatically generating image captions. However, evaluation is challenging. Existing automatic evaluation metrics are primarily sensitive to n-gram overlap, which is neither necessary nor sufficient for the task of simulating human judgment. We hypothesize that semantic propositional content is an important component of human caption evaluation, and propose a new automated caption evaluation metric defined over scene graphs coined SPICE. Extensive evaluations across a range of models and datasets indicate that SPICE captures human judgments over model-generated captions better than other automatic metrics (e.g., system-level correlation of 0.88 with human judgments on the MS COCO dataset, versus 0.43 for CIDEr and 0.53 for METEOR). Furthermore, SPICE can answer questions such as which caption-generator best understands colors? and can caption-generators count?

1,053 citations

Journal ArticleDOI
TL;DR: Differences in heterozygosity indicate lowered evolutionary potential, compromised reproductive fitness, and elevated extinction risk in the wild in threatened species.
Abstract: There is controversy concerning the role of genetic factors in species extinctions. Many authors have asserted that species are usually driven to extinction before genetic factors have time to impact them, but few studies have seriously addressed this issue. If this assertion is true, there will be little difference in genetic diversity between threatened and taxonomically related nonthreatened species. We compared average heterozygosities in 170 threatened taxa with those in taxonomically related nonthreatened taxa in a comprehensive metaanalysis. Heterozygosity was lower in threatened taxa in 77% of comparisons, a highly significant departure from the predictions of the no genetic impact hypothesis. Heterozygosity was on average 35% lower (median 40%) in threatened taxa than in related nonthreatened ones. These differences in heterozygosity indicate lowered evolutionary potential, compromised reproductive fitness, and elevated extinction risk in the wild. Independent evidence from stochastic computer projections has demonstrated that inbreeding depression elevates extinction risk for threatened species in natural habitats when all other threatening processes are included in the models. Thus, most taxa are not driven to extinction before genetic factors affect them adversely.

1,049 citations

Journal ArticleDOI
01 May 1990-Pain
TL;DR: Overall, the faces pain scale incorporates conventions used by children, has achieved strong agreement in the rank ordering of pain, has indications that the intervals are close to equal, and is treated by children as a scale.
Abstract: Altogether 553 children (195 first graders, mean age 6.8 years, and 358 third graders, mean age 8.7 years) participated in the development of a self-report measure to assess the intensity of children's pain. The first step was the derivation, from children's drawings of facial expressions of pain, of 5 sets of 7 schematic faces depicting changes in severity of expressed pain from no pain to the most pain possible. With the set of faces that achieved the highest agreement in pain ordering, additional studies were conducted to determine whether the set had the properties of a scale. In one study, children rank-ordered the faces on 2 occasions, separated by 1 week. All 7 faces were correctly ranked by 64% (retest 1 week later, 61%) of grade 1 children and by 86% (retest 89%) of grade 3 children. In a second study, the faces were presented in all possible paired combinations. All 7 faces were correctly placed by 62% (retest 86%) of the younger and by 75% (retest 71%) of the older subjects. A third study asked children to place faces along a scale: a procedure allowing a check on the equality of intervals. The fourth study checked on whether pain was acting as an underlying construct for ordering the faces in memory. We asked whether children perceived the set as a scale by asking if memory for an ordered set of faces was more accurate than for a random set. The final study checked, with 6-year-old children, the test-retest reliability of ratings for recalled experiences of pain. Overall, the faces pain scale incorporates conventions used by children, has achieved strong agreement in the rank ordering of pain, has indications that the intervals are close to equal, and is treated by children as a scale. The test-retest data suggest that it may prove to be a reliable index over time of self-reported pain.

1,031 citations

Posted Content
TL;DR: An interpretive framework is presented that analyzes the definitional perspectives and the applications of big data, and a general taxonomy is provided that helps broaden the understanding ofbig data and its role in capturing business value.
Abstract: Big data has the potential to revolutionize the art of management. Despite the high operational and strategic impacts, there is a paucity of empirical research to assess the business value of big data. Drawing on a systematic review and case study findings, this paper presents an interpretive framework that analyzes the definitional perspectives and the applications of big data. The paper also provides a general taxonomy that helps broaden the understanding of big data and its role in capturing business value. The synthesis of the diverse concepts within the literature on big data provides deeper insights into achieving value through big data strategy and implementation.

1,024 citations

Journal ArticleDOI
TL;DR: It is shown that parametric down-conversion offers enhanced performance compared to its weak coherent pulse counterpart and existing experimental schemes (based on weak pulses) currently do not offer unconditional security for the reported distances and signal strength.
Abstract: We provide limits to practical quantum key distribution, taking into account channel losses, a realistic detection process, and imperfections in the ``qubits'' sent from the sender to the receiver. As we show, even quantum key distribution with perfect qubits might not be achievable over long distances when the other imperfections are taken into account. Furthermore, existing experimental schemes (based on weak pulses) currently do not offer unconditional security for the reported distances and signal strength. Finally we show that parametric down-conversion offers enhanced performance compared to its weak coherent pulse counterpart.

1,020 citations


Authors

Showing all 14346 results

NameH-indexPapersCitations
Yang Yang1712644153049
Peter B. Reich159790110377
Nicholas J. Talley158157190197
John R. Hodges14981282709
Thomas J. Smith1401775113919
Andrew G. Clark140823123333
Joss Bland-Hawthorn136111477593
John F. Thompson132142095894
Xin Wang121150364930
William L. Griffin11786261494
Richard Shine115109656544
Ian T. Paulsen11235469460
Jianjun Liu112104071032
Douglas R. MacFarlane11086454236
Richard A. Bryant10976943971
Network Information
Related Institutions (5)
Australian National University
109.2K papers, 4.3M citations

96% related

University of Queensland
155.7K papers, 5.7M citations

96% related

University of Sydney
187.3K papers, 6.1M citations

95% related

University of Melbourne
174.8K papers, 6.3M citations

95% related

University of New South Wales
153.6K papers, 4.8M citations

94% related

Performance
Metrics
No. of papers from the Institution in previous years
YearPapers
2023110
2022463
20214,106
20204,009
20193,549
20183,119