Catalogue Search | MBRL
Are you sure you want to remove the book from the shelf?
{{itemTitle}}
14,630
result(s) for
"Williams, Christopher"
Sort by:
The Pascal Visual Object Classes Challenge: A Retrospective
The
Pascal
Visual Object Classes (VOC) challenge consists of two components: (i) a publicly available
dataset
of images together with ground truth annotation and standardised evaluation software; and (ii) an annual
competition
and workshop. There are five challenges: classification, detection, segmentation, action classification, and person layout. In this paper we provide a review of the challenge from 2008–2012. The paper is intended for two audiences:
algorithm designers
, researchers who want to see what the state of the art is, as measured by performance on the VOC datasets, along with the limitations and weak points of the current generation of algorithms; and,
challenge designers
, who want to see what we as organisers have learnt from the process and our recommendations for the organisation of future challenges. To analyse the performance of submitted algorithms on the VOC datasets we introduce a number of novel evaluation methods: a bootstrapping method for determining whether differences in the performance of two algorithms are significant or not; a normalised average precision so that performance can be compared across classes with different proportions of positive instances; a clustering method for visualising the performance across multiple algorithms so that the hard and easy images can be identified; and the use of a joint classifier over the submitted algorithms in order to measure their complementarity and combined performance. We also analyse the community’s progress through time using the methods of Hoiem et al. (Proceedings of European Conference on Computer Vision,
2012
) to identify the types of occurring errors. We conclude the paper with an appraisal of the aspects of the challenge that worked well, and those that could be improved in future challenges.
Journal Article
Interventions to reduce social isolation and loneliness during COVID-19 physical distancing measures: A rapid systematic review
A significant proportion of the worldwide population is at risk of social isolation and loneliness as a result of the COVID-19 pandemic. We aimed to identify effective interventions to reduce social isolation and loneliness that are compatible with COVID-19 shielding and social distancing measures.
In this rapid systematic review, we searched six electronic databases (Medline, Embase, Web of Science, PsycINFO, Cochrane Database of Systematic Reviews and SCOPUS) from inception to April 2020 for systematic reviews appraising interventions for loneliness and/or social isolation. Primary studies from those reviews were eligible if they included: 1) participants in a non-hospital setting; 2) interventions to reduce social isolation and/or loneliness that would be feasible during COVID-19 shielding measures; 3) a relevant control group; and 4) quantitative measures of social isolation, social support or loneliness. At least two authors independently screened studies, extracted data, and assessed risk of bias using the Downs and Black checklist. Study registration: PROSPERO CRD42020178654. We identified 45 RCTs and 13 non-randomised controlled trials; none were conducted during the COVID-19 pandemic. The nature, type, and potential effectiveness of interventions varied greatly. Effective interventions for loneliness include psychological therapies such as mindfulness, lessons on friendship, robotic pets, and social facilitation software. Few interventions improved social isolation. Overall, 37 of 58 studies were of \"Fair\" quality, as measured by the Downs & Black checklist. The main study limitations identified were the inclusion of studies of variable quality; the applicability of our findings to the entire population; and the current poor understanding of the types of loneliness and isolation experienced by different groups affected by the COVID-19 pandemic.
Many effective interventions involved cognitive or educational components, or facilitated communication between peers. These interventions may require minor modifications to align with COVID-19 shielding/social distancing measures. Future high-quality randomised controlled trials conducted under shielding/social distancing constraints are urgently needed.
Journal Article
The Pascal Visual Object Classes (VOC) Challenge
The
Pascal
Visual Object Classes (VOC) challenge is a benchmark in visual object category recognition and detection, providing the vision and machine learning communities with a standard dataset of images and annotation, and standard evaluation procedures. Organised annually from 2005 to present, the challenge and its associated dataset has become accepted as
the
benchmark for object detection.
This paper describes the dataset and evaluation procedure. We review the state-of-the-art in evaluated methods for both classification and detection, analyse whether the methods are statistically different, what they are learning from the images (e.g. the object or its context), and what the methods find easy or confuse. The paper concludes with lessons learnt in the three year history of the challenge, and proposes directions for future improvement and extension.
Journal Article
Home at last
\"After Lester is adopted from his foster home by Daddy Albert and Daddy Rich, he can't fall asleep in his new bed. What will it take to make Lester feel home at last?\" -- Provided by publisher.
Book
Evaluating the use of large language models to provide clinical recommendations in the Emergency Department
The release of GPT-4 and other large language models (LLMs) has the potential to transform healthcare. However, existing research evaluating LLM performance on real-world clinical notes is limited. Here, we conduct a highly-powered study to determine whether LLMs can provide clinical recommendations for three tasks (admission status, radiological investigation(s) request status, and antibiotic prescription status) using clinical notes from the Emergency Department. We randomly selected 10,000 Emergency Department visits to evaluate the accuracy of zero-shot, GPT-3.5-turbo- and GPT-4-turbo-generated clinical recommendations across four different prompting strategies. We found that both GPT-4-turbo and GPT-3.5-turbo performed poorly compared to a resident physician, with accuracy scores 8% and 24%, respectively, lower than physician on average. Both LLMs tended to be overly cautious in its recommendations, with high sensitivity at the cost of specificity. Our findings demonstrate that, while early evaluations of the clinical use of LLMs are promising, LLM performance must be significantly improved before their deployment as decision support systems for clinical recommendations and other complex tasks.
The emergence of large language models has the potential to transform healthcare. Here, the authors show that, when providing clinical recommendations, these models perform poorly compared to physicians and are overly cautious in their decisions.
Journal Article
Asymmetric Responses to Earnings News: A Case for Ambiguity
This study empirically examines the role of shocks to macro-uncertainty in shaping the responses of stock market participants to firm-specific earnings news. Specifically, I find that investors place greater weight on bad news following an increase in macro-uncertainty. By contrast, I find that investors place equal weight on both good and bad news following a decrease in macro-uncertainty. Furthermore, my findings show that these effects are more pronounced (1) for firms whose prior returns are more correlated with macro-uncertainty, (2) for firms that experience abnormally low trading volume during the earnings announcement, (3) for firms with relatively lower levels of institutional ownership, and (4) for firms with relatively higher information uncertainty. In sum, these findings provide novel empirical evidence that investors behave in a manner consistent with ambiguity aversion, with the effects strongest among unsophisticated investors.
Journal Article