Conference / Webinar
In This Talk, Ayodele Odubela covers how to see the bias in your data and machine learning models. By giving examples from recent research on biased facial recognition systems as well as cutting edge explainability algorithms, viewers can expect clarification on what bias means in machine learning.
Ayodele Odubela is a Data Scientist working on driver risk mitigation at SambaSafety in Denver, CO. Previously she earned her Master’s degree in Data Science after transitioning to tech from social media marketing. She’s created algorithms that predict consumer segment movement, goals in hockey, and the location of firearms using radio frequency sensors.
Juneteenth Conference is a free virtual tech conference made for and featuring Black people in Technology. The conference celebrates Black Excellence and promotes community for Black people who are severely underrepresented, overlooked, and underutilized in the tech industry.
Key Takeaways
1. Biased data = biased outcomes
Data used for training models is hardly ever actually representative of people it will be used on
2. Humans are imperfect,
This doesn’t mean algorithms are unbiased, it means we have to assume bias will persist until we take steps to remove it
3. Master of our demise
Data science is one of the few fields where professionals get to pick their success metric
