Maximizing Model Accuracy With Data-centric AI Practices


Feb 08, 05:00PM PDT(01:00AM GMT).
  • Free 84 Attendees
Description
Speaker
This event is hosted by SF Big Analytics Group. https://www.meetup.com/SF-Big-Analytics

In computer vision projects, managing data is often an afterthought. That is why at least 31% of projects fail. In this webinar, the speaker will talk about how to build a solid data foundation to train more accurate and cost-effective computer vision models.

In this session, you will learn about Hub, an open-source dataset format for AI. Hub works with computer vision datasets of any size and enables easy creation, storage, version control, and streaming to ML frameworks while training. Moreover, you will learn how to apply the data-centric framework to resolve common data bottlenecks when using tools like Amazon SageMaker.

The speaker will also demonstrate how to visualize and explore datasets – from MNIST to ImageNet. As a result of the session, you will be able to easily build computer vision data pipelines and fully utilize the compute resources.

Davit Buniatyan

The idea of creating Hub came to Davit while he was pursuing a Ph.D. at Princeton Neuroscience Lab, where his research involved reconstructing the topology of mouse brains. As part of his research, he dealt with large-scale unstructured data which could cost millions of dollars a year to store. That inspired Davit to create Hub, and eventually Activeloop, the database for AI company.
The event ended.
Watch Recording
*Recordings hosted on Youtube, click the link will open the Youtube page.