Project Polar- Energy efficient artificial intelligence using photonic processors

Project Polar——使用光子处理器的节能人工智能

• 批准号: 86749
• 金额: $ 12.74万
• 依托单位: ORCA COMPUTING LTD.
• 依托单位国家: 英国
• 项目类别: Collaborative R&D
• 财政年份: 2020
• 资助国家: 英国
• 起止时间: 2020 至 无数据
• 项目状态: 已结题
• 来源: https://gtr.ukri.org/projects?ref=86749
• 关键词: Project; Polar; Energy; efficient; artificial

While the rise of artificial intelligence and big data are often presented as the answer to the challenge of big data, they also present a significant challenge to the environment. As the size and complexity of AI algoritms grows, the energy consumption and environmental cost related to training those algorithms and analysing data is growing at an alarming rate. For example, the average cost of training an AI algorithm is as much as 284 tonnes, five times the emissions of an average car, and the amount of computing time and energy needed has doubled every 3.4 months since 2012\.ORCA computing are a brand new photonic computing company, spun out from the University of Oxford in November 2019\. In this project, ORCA will develop a prototype photonic integrated processor system that can enable significant speed up and efficiency savings over the AI accelerator products on the market today.

虽然人工智能和大数据的兴起通常被认为是对大数据挑战的答案，但它们也对环境提出了重大挑战。随着人工智能算法的规模和复杂性不断增长，与训练这些算法和分析数据相关的能源消耗和环境成本正以惊人的速度增长。例如，训练一个人工智能算法的平均成本高达 284 吨，是普通汽车排放量的五倍，自 2012 年以来，所需的计算时间和能源每 3.4 个月就翻一番\。ORCA 计算是一家全新的光子计算公司，于 2019 年 11 月从牛津大学分拆出来\。在该项目中，ORCA 将开发原型光子集成处理器系统，与当今市场上的人工智能加速器产品相比，该系统可以显着提高速度并节省效率。