We invite all confirmed Deep Learning Indaba attendees to apply to present their research as a spotlight talk at the Machine Learning Efficiency workshop on Friday, August 26th, 2022.
๐ช๐ผ๐ฟ๐ธ๐๐ต๐ผ๐ฝ ๐ฑ๐ฒ๐๐ฐ๐ฟ๐ถ๐ฝ๐๐ถ๐ผ๐ป
Most of the world is using machine learning in resource-constrained environments. In contrast, modern machine learning models are becoming larger, with computationally intensive training procedures.The rise of these larger models have also led to efforts to make machine learning more efficient and deployable at the edge. In this workshop, we put a focus on efficiency research and ML deployment at the edge in Africa.
๐ช๐ต๐ฎ๐ ๐๐๐ฝ๐ฒ ๐ผ๐ณ ๐๐ผ๐ฟ๐ธ ๐ถ๐ ๐ฎ ๐ฐ๐ฎ๐ป๐ฑ๐ถ๐ฑ๐ฎ๐๐ฒ ๐ณ๐ผ๐ฟ ๐ฎ ๐๐ฝ๐ผ๐๐น๐ถ๐ด๐ต๐ ๐๐ฎ๐น๐ธ ?
We are interested in work that discusses the challenges and opportunities for using machine learning in resource constrained environments. We define resource constrained broadly to refer to communities with limited resources required for machine learning training (limited compute) ย and/or inference (limited internet connectivity, high sensitivity to data prices or low bandwidth). ย The work can address an applied problem (discussion of challenges and opportunities deploying models to resource constrained environments), general solutions to improve the efficiency of models (e.g. more efficient data sampling, algorithmic solutions like model pruning, quantization, federated learning, hardware solutions, IOT, efficient data sampling) or an open source engineering contribution (e.g. releasing open source code that helps the general community in some way).
๐๐๐ญ๐๐ข๐ฅ๐ฌ
Machine Learning Efficiency Workshop
Date: Friday August 26th 2022
Time: 8.30am - 5pm
๐ฆ๐ฝ๐ผ๐๐น๐ถ๐ด๐ต๐ ๐๐ฎ๐น๐ธ๐
Successful applicants will each give a 8 minute oral presentation (with slides) followed by 2 minutes of questions from the audience. When making your slides, remember that you will only have 8 minutes to present, so less than 10 and at most 15 slides are recommended. Slides can be in Microsoft Powerpoint, a link to Google slides or a PDF of the slides.
As a speaker, you will have the opportunity to have your work profiled and engage with fellow researchers. ย Need more guidance? See some of the spotlight talks profiled in the previous ml at the edge workshop:
https://drive.google.com/file/d/1amEsiqnBkH7ZUMtwK900TEBLPWFyWXCZ/view?usp=sharing.
