Reducing AI Bias and Optimizing Data Labeling Frameworks w/ Appen

Logo
Presented by

Monchu Chen, Principal Data Scientist @ Appen

About this talk

Tentative Schedule: 2:00pm: Intro 2:05pm: Reducing AI Bias and Optimizing Data Labeling Frameworks w/ Appen by Monchu Chen 2:45pm: Q&A Talk Abstract: Bias in machine learning has become a significant concern as AI technology spreads to more application domains.  While some bias is a consequence of limits in design and tooling, bias in the training data itself is much more common. Skewed training data often promotes AI models that reveal discrimination and amplify human prejudices. In this talk, we present a framework, developed at Appen, to minimize bias. This framework operates by routing data labeling tasks to the right labelers to avoid introducing bias. It also optimizes the process by determining a proper distribution of labelers for a given task. Our speaker, Monchu Chen, will review some use cases where this framework has been applied, and discuss results that show how the optimization process minimizes skew in the training data. Chen will also discuss extending this approach to other use cases and review the implications of this work. Speaker bios: Monchu Chen has worked in human-computer interaction for more than two decades.  He has helped corporations and startups apply user insights to product innovation in multiple application domains. Monchu now focuses on the human aspects of AI as the Principal Data Scientist for Appen's ML team, building models and systems to improve annotation quality, efficiency, and reducing AI bias. Dr. Chen holds a PhD from Carnegie Mellon University. He previously held a tenured faculty position at Carnegie Mellon Portugal and is the author of more than 70 publications and patents.
Related topics:

More from this channel

Upcoming talks (0)
On-demand talks (265)
Subscribers (55708)
Dataiku is the world’s leading platform for Everyday AI, systemizing the use of data for exceptional business results. Organizations that use Dataiku elevate their people (whether technical and working in code or on the business side and low- or no-code) to extraordinary, arming them with the ability to make better day-to-day decisions with data. More than 450 companies worldwide use Dataiku to systemize their use of data and AI, driving diverse use cases from fraud detection to customer churn prevention, predictive maintenance to supply chain optimization, and everything in between.