AI DATA ACCURACY - How To Approach Working With Datasets with Jason Corso, Co-Founder of Voxel51, Part 1

Apr 11, 2024

Ask episode

Chapters

Transcript

Episode notes

Today’s episode welcomes Jason Corso - Professor of Robotics and EECS at University of Michigan and Co-Founder/Chief Science Officer at Voxel51 - a Series A AI software company that enables machine learning and computer vision scientists to rapidly curate and experiment with their datasets in order to build higher performing machine learning systems.

Jason is a builder by nature. During his academic career he started to ask himself - “can I have a bigger impact on the world?”, which led him to meet his co-founder and start Voxel 51. Both of them had a passion for machine learning and an ability to write code - which is how his transition from academia to consulting began. In the early days they received a grant which worked out well and funded them through the process of building their first service.

Building a service, however, was never part of the plan. Once they realized that the limitations of the grant won’t enable building a product, they sought out venture capital. This venture capital boost allowed them to build out their open-source platform which has more than 2 million downloads as of this episode.

Jason speaks on the early success stories of Voxel51. One of their earliest clients took just 1% of their data and in a few hours they were able to look at the data visually, which enabled them to figure out where mistakes were made. Data quality is more important than the sophistication of the system, and Voxel51 enables their clients to find where inefficiencies live.

Even with a great system - human input is still needed when it comes to quality control. Jason speaks on how humans filling in the gaps when the system is less certain is highly effective - even though it doesn’t scale. Eric speaks on how experienced data scientists develop an almost 6th sense when looking at data and understanding that something is wrong and a change is in order.

Check out the full episode to learn more on how to approach utilizing data!

HIGHLIGHTS:
03:41 Building a service-based business
05:21 When do you utilize Voxel51 in the development process?
09:52 Quality or quantity - what’s more important for your data?
15:05 The importance of human pattern recognition
18:03 New applications of AI tools

Connect with Jason - https://www.linkedin.com/in/jason-corso/

Check out Voxel51 - https://voxel51.com/

Don't forget to subscribe to the Chaos to Clarity Podcast for more invaluable episodes to help you grow your business and stay ahead of the curve!To reach out to Eric, visit https://chaostoclarity.io/