Facial recognition networks have demonstrated disproportionately poor results in identifying women or people of color. The use of discriminatory models leads to results that may be racist, sexist, or biased in other ways.
At the Embedded Vision Summit, Perceive CEO Steve Teig spoke about the issues around poor statistical reasoning that contribute to these biased results, and explored how models can be trained in ways that can mitigate bias.
For more, please visit the Edge AI + Vision Alliance website here.