No-Code Machine Learning Tool Unveiled
- By David Ramel
- September 6, 2018
Low-code and even no-code tools have increased as enterprises seek to quickly ship apps amid a dearth of programming talent, but they have mostly targeted simple projects -- until now. The drag-and-drop approach has now been put to use for a much more complicated task in the artificial intelligence space: creating machine learning models.
The tool is called EZDL, and it was recently unveiled by Chinese AI specialist Baidu.
EZDL employs a four-step process for project development and deployment of custom ML models: create a model; upload and label images/objects; train and test the model; and deploy it with a cloud API or offline SDK.
The company said the tool targets small and medium-size enterprises.
"Even if you have had no exposure to programming, you can quickly build models on this platform with zero barriers," Yongkang Xie, tech lead of Baidu EZDL, said in a statement. "EZDL can help companies with limited AI experts and computing resources to quickly and efficiently complete deep learning training and deployment with only a small amount of data."
Baidu is highlighting use cases in three categories:
- An image classification model for the automatic classification of images by custom label, for tasks such as: classification of home decorating images; image recognition of Chinese herbal medicine, wild birds and frogs; industrial quality inspection to identify defective products; and pavement damage monitoring.
- An object detection model for the automatic detection of objects in images and counting the number of objects by label, for tasks such as retail inspection and medical detection for cell counting.
- A sound classification model for recognizing types of sounds or detecting classes of a condition/event, for tasks such as security monitoring and scientific research.
The company said the new tool is a furthering of its effort to democratize access to AI capabilities, building on its Baidu Brain offerings -- a full AI tech stack including chips, deep learning frameworks and platforms -- which recently shipped in version 3.0 after debuting two years ago.
More information on the new low-code tool can be found in a FAQ.
David Ramel is an editor and writer for Converge360.