5,809 research outputs found
Do dolphins benefit from nonlinear mathematics when processing their sonar returns?
No full textAn interview with author Tim Leighton about the paper
Opportunities for linking young surveyors across professional surveying member organisations and FIG
No full textGreen AI: An empirical study
No full textIn this work, we look at the intersection of Sustainable Software Engineering and AI engineering known as Green AI. AI computing is rapidly becoming more expensive, calling for a change in design philosophy. We consider both training and inference of neural networks used for image vision; to reveal energy-efficient practices in an exploratory fashion.First of all, we examine a modern algorithm for hyperparameter optimisation and compare this to two baseline methods. We find that the baseline algorithms perform considerably worse despite their wide usage and argue that they should not be used when training large models. Furthermore, we look at the layer structure of convolutional networks and conclude that the convolutional layers have the largest influence on the total consumption. We report increases of up to 95% with only marginal improvements in accuracy. Therefore we recommend developers to reduce their network architectures as long as the performance stays within a reasonable margin. Second, we present a study focused on the inference phase of the deep learning pipeline. We look at the effect of batching for image classification requests. To facilitate the data collection, we make use of a simulated queue and the Pytorch framework. We find that batching has a significant impact on the energy consumption, but the magnitude of this impact can vary a lot for different models. Our recommendation is to treat the batch size as an inference parameter that needs to be tuned first. Additionally, we highlight how the energy consumption of image vision networks has evolved over the past decade. Presenting the findings together with the performance of these networks shows a steady, upward energy trend accompanied by a decreasing slope for the accuracy. The only exception is the model ShuffleNetV2. We mention the design principles that went into the development of this network and present it as a start for future research.Computer Scienc
Tim Di Muzio on 'Sabotage'
No full textIn a series of essays published in 2013 and 2014 on capitaspower.com, political economist Tim Di Muzio explored the concept of ‘sabotage’ as it applies to capitalist power. I recently rediscovered these essays and was so impressed by them that I have reposted them here as a single piece.
About the author: Tim Di Muzio is a researcher at the University of Wollongong. He is the author of numerous books, including Debt as power, Carbon capitalism, and The 1% and the Rest of us
1996-1997 Tim Gautreaux
No full textTim Gautreaux is the author of three novels and two earlier short story collections. His work has appeared in The New Yorker, The Best American Short Stories, The Atlantic, Harper’s, and GQ. After teaching for thirty years at Southeastern Louisiana University, he now lives, with his wife, in Chattanooga, Tennessee. (Photo credit: Randy Bergeron)https://egrove.olemiss.edu/grisham_res/1023/thumbnail.jp
Material of the paper entitled "Uncovering Energy-Efficient Practices in Deep Learning Training: Preliminary Steps Towards Green AI"
No full textWe provide the scripts used to carry out the experiments, as well as the associated generated datasets, described in the paper entitled "Uncovering Energy-Efficient Practices in Deep Learning Training: Preliminary Steps Towards Green AI". The material can be used to replicate the experiments and results.
Yarally, T., Cruz, L., Feitosa, D., Sallou, J., van Deursen, A. (2023). Uncovering Energy-Efficient Practices in Deep Learning Training: Preliminary Steps Towards Green AI
First person - Tim Petzold
No full textFirst Person is a series of interviews with the first authors of a selection of papers published in Biology Open, helping researchers promote themselves alongside their papers. Tim Petzold is first author on ‘ Connexin 41.8 governs timely haematopoietic stem and progenitor cell specification’, published in BiO. Tim conducted the research described in this article while a PhD student in Julien Bertrand's lab at the Department of Pathology and Immunology, Faculty of Medicine, University of Geneva, Switzerland. He is now a postdoc in the lab of Holger Gerhardt at the Max Delbrück Center for Molecular Medicine in the Helmholtz Association, Berlin, Germany, investigating developmental biology – previously his focus was on how blood stem cells develop and now it has shifted to how the vascular system develops
Tim Seibles, 40th Annual ODU Literary Festival
No full textTim Seibles is the author of several poetry collections including Hurdy-Gurdy, Hammerlock, Buffalo Head Solos, and Fast Animal, which was a finalist for the 2012 National Book Award. In 2013 he received both the Pen Oakland Josephine Miles Award for poetry and an honorary Doctorate of Humane Letters from Misericordia University for his literary accomplishments. His latest collection, One Turn Around the Sun, has just been released. Tim is the current Poet Laureate of Virginia and is a Professor of English at Old Dominion University where he teaches literature as well as classes in the MFA in writing program
Tim Seibles, 39th Annual ODU Literary Festival
No full textTim Seibles is the author of several poetry collections including Hurdy-Gurdy, Hammerlock, Buffalo Head Solos, and Fast Animal, which was a finalist for the 2012 National Book Award. In 2013 he received both the Pen Oakland Josephine Miles Award for poetry and an honorary Doctorate of Humane Letters from Misericordia University for his literary accomplishments. His latest collection, One Turn Around the Sun, has just been released. Tim is the current Poet Laureate of Virginia and is a Professor of English at Old Dominion University where he teaches literature as well as classes in the MFA in writing program
Global Media Ideas - Infinite Pathways to Creative Succes - Tim Chang - Part One.mp4
No full textDuring the X Media Lab: Global Media Ideas summit in June 2011 media and technology writer Brad Howarth conducted interviews with industry experts for Creativeinnovation. This video is part one of Brad Howarth's interview with Tim Chang about his role as Partner at Norwest Venture Partners (Palo Alto). Tim focuses on investments in mobile, gaming, digital media, and also leads Norwest Venture Partners's investment practice in China and Asia-Pacific. Tim shares tips on how to get an introduction to a Venture Capital; the elements of a good pitch and follow-up. And what he looks at when considering a deal - The 3Ts: Team, Traction, Tier 1 co-investors
- …
