李飞飞宣布谷歌云里程碑事件:推出Cloud AutoML
本文系网易智能工作室(公众号 )出品。聚焦AI,读懂下一个大时代!
【网易智能讯1月18日消息】李飞飞在推特表示“在短短的几个月里,将尖端技术转化为数百万的产品,这是一个相当鼓舞人心的旅程!我们希望 是我们客户的第一选择。”
李飞飞截图
据了解,谷歌Cloud 系统基于监督学习,所以需要提供一系列带有标签的数据。具体来说,开发者只需要上传一组图片,然后导入标签或者通过App创建,随后谷歌的系统就会自动生成一个定制化的机器学习模型。
值得一提的是,Cloud 是个开发者的利器,即便你不懂机器学习,也能训练出一个定制化的机器学习模型。
据悉,目前谷歌还没有公布Cloud 的服务价格,开发者还需要等待。(定西)
【附:李飞飞博客全文】
Cloud : AI to every
When we both Cloud just over a year ago, we on a to AI. Our goal was to lower the of entry and make AI to the of , and .
Our Cloud AI team has been good this goal. In 2017, we Cloud , to help with build ML that work on any type of data, of any size. We how , i.e., APIs— , , NLP, and —could be built upon pre- to bring scale and speed to . , our of data and ML , has grown to more than one . And today, more than 10,000 are using Cloud AI , like Box, Rolls Royce , and Ocado.
But there’s much more we can do. , only a of in the world have to the and to fully the of ML and AI. There’s a very of that can . And if you’re one of the that has to ML/AI , you still have to the time- and of your own ML model. While has pre- via APIs that tasks, there's still a long road ahead if we want to bring AI to .
To close this gap, and to make AI to every , we’re Cloud . Cloud helps with ML start their own high- by using like and from . We Cloud will make AI even more , new in AI and help less- build AI they only of.
Our first Cloud will be Cloud , a that makes it and to ML for image . Its drag-and-drop lets you , train and , and then those on Cloud. Early using Cloud to like and CIFAR have shown more with fewer than ML APIs.
Here’s a more on what Cloud has to offer:
: Cloud is built on ’s image , and . This means you’ll get a more model even if your has .
time to -ready : With Cloud , you can a model in to pilot your AI- , or build out a full, -ready model in as as a day.
Easy to use: a user that lets you data, then turns that data into a high model for your needs.
“Urban is for new ways to our ’ ," says Alan , Data at URBN. " and a set of is to our , and ; , is and time-. To this, our team has been Cloud to the by like and . Cloud has great to help our with , and ."
Mike White, CTO and SVP, for and Media, says: “Cloud ’s is us build to our with , and . These are being into our to the on Guest more , and on .”
And , Lead at the of , tells us: "ZSL is an to the of and their 。 A key to on this is to track to learn more about their and the are on these 。 In order to this, ZSL has a of traps in the wild that take of when by heat or 。
The of by these are then and with the , such as , lions and , etc。, which is a - and 。 ZSL’s Unit has been with ’s Cloud ML team to help shape the of this , which ZSL aims to use to the of these — costs, wider-scale and a of how to the world’s 。"
If you’re in out , you can via this form.
is the of our close with Brain and other AI teams, and is the first of Cloud in . While we’re still at the of our to make AI more , we’ve been by what our 10,000+ using Cloud AI have been able to . We hope the of Cloud will help even more what’s AI.