Hubs
Live
My Hubs
All
My Hubs
All

Trending
Today
Trending
Today

# Sign in to discover all of the research papers you care about, live as they're published.

70
Authors: Amanda J Ullman, Patricia M Davidson
Authors: Amanda J Ullman, Patricia M Davidson
46
Authors: Ng, Victoria, et al
Authors: Ng, Victoria, et al
Background: Shutdowns are enacted when alternative public health measures are insufficient to control the epidemic and the population is largely susceptible. An age-stratified agent-based model was developed to explore the impact of shutdowns to control SARS-CoV-2 transmission in Canada under the assumption that current efforts to control the epidemic remains insufficient and in the absence of a vaccine. Methods: We estimated the current levels of interventions in Canada to generate a baseline scenario from 7 February to 7 September 2020. Four aspects of shutdowns were explored in scenarios that ran from 8 September 2020 to 7 January 2022, these included the impact of how quickly shutdowns are implemented, the duration of shutdowns, the minimum break (delays) between shutdowns and the types of sectors to shutdown. Comparisons among scenarios were made using cases, hospitalizations, deaths and shutdown days during the 700-day model runs. Results: We found a negative relationship between reducing SARS-CoV-2 transmission and the number of shutdown days. However, we also found that for shutdowns to be optimally effective, they need to be implemented fast with minimal delay, initiated when community transmission is low, sustained for an adequate period and be stringent and target multiple sectors, particularly those driving transmission. By applying shutdowns in this manner, the total number of shutdown days could be reduced compared to delaying the shutdowns until further into the epidemic when transmission is higher and/or implementing short insufficient shutdowns that would require frequent re-implementation. This paper contrasts a range of shutdown strategies and trade-offs between health outcomes and economic metrics that need to be considered within the local context. Interpretation: Given the immense socioeconomic impact of shutdowns, they should be avoided where possible and used only when other public health measures are insufficient to control the epidemic. If used, the time it buys to delay the epidemic should be used to enhance other equally effective, but less disruptive, public health measures.
99
Authors: Alex Nichol, Prafulla Dhariwal, Alex Nichol
Authors: Alex Nichol, Prafulla Dhariwal, Alex Nichol
We show that diffusion models can achieve image sample quality superior to the current state-of-the-art generative models. We achieve this on unconditional image synthesis by finding a better architecture through a series of ablations. For conditional image synthesis, we further improve sample quality with classifier guidance: a simple, compute-efficient method for trading off diversity for sample quality using gradients from a classifier. We achieve an FID of 2.97 on ImageNet $128 \times 128$, 4.59 on ImageNet $256 \times 256$, and $7.72$ on ImageNet $512 \times 512$, and we match BigGAN-deep even with as few as 25 forward passes per sample, all while maintaining better coverage of the distribution. Finally, we find that classifier guidance combines well with upsampling diffusion models, further improving FID to 3.85 on ImageNet $512 \times 512$. We release our code at https://github.com/openai/guided-diffusion
Slide 1 of 1
50
Authors: Yousefzadeh, Matthew J., et al
Authors: Yousefzadeh, Matthew J., et al
57
Authors: Amy Maxmen
Authors: Amy Maxmen
32
Authors: David Arnold, Will Dobbie, Peter Hull
Authors: David Arnold, Will Dobbie, Peter Hull
Algorithmic decision-making can lead to discrimination against legally protected groups, but measuring such discrimination is often hampered by a fundamental selection challenge. We develop new quasi-experimental tools to overcome this challenge and measure algorithmic discrimination in pretrial bail decisions. We show that the selection challenge reduces to the challenge of measuring four moments, which can be estimated by extrapolating quasi-experimental variation across as-good-as-randomly assigned decision-makers. Estimates from New York City show that both a sophisticated machine learning algorithm and a simpler regression model discriminate against Black defendants even though defendant race and ethnicity are not included in the training data.
98
Authors: Emily Waltz
Authors: Emily Waltz
49
Authors: Helen Pearson