))
Is A.I. better at avoiding bias?
Manage episode 343942146 series 3238641
Contenu fourni par Anchormen. Tout le contenu du podcast, y compris les épisodes, les graphiques et les descriptions de podcast, est téléchargé et fourni directement par Anchormen ou son partenaire de plateforme de podcast. Si vous pensez que quelqu'un utilise votre œuvre protégée sans votre autorisation, vous pouvez suivre le processus décrit ici https://fr.player.fm/legal.
This podcast takes off with Jeroen and Ron talking about how algorithms can become biased and they discuss this on the basis of the gender bias hiring example. How can you avoid black box algorithms and force the neural network to represent its decision making process?
Next, they touch upon the accuracy of face and emotion recognition and how this relates to the 'dream' of Artificial General Intelligence (AGI). Can machines actually point into places where humans didn't go yet? (Spoiler: AlphaGo Zero)
What can companies learn from this: who takes the responsibility to avoid bias and to have a balanced, unbiased data (training) set? Jeroen and Ron explain why Precision and Recall are better metrics (over accuracy) to check whether your algorithm or data set is unbiased or not. And how can recommendation engines combined with post-processing help avoid collaborative filtering.
8 episodes
Partager
