Sure, it is easier to swap algorithms than to change human minds, but the rest of this @pmdomingos argument is flawed.
Here's a mini-thread on why.
(1/6)https://twitter.com/pmddomingos/status/1383958310102061061 …
Näytä tämä ketju
1. On the contrary, it's the alleged demonstrations of past bias you list that are flawed. 2. Learning from past data does *not* imply repeating the mistakes of the past. This is a (unfortunately pervasive) confusion between the learning system and the decision system.
-
-
So you want to replace the predictive learning targets by an invented future of your liking, perhaps influenced by utopian ideas? I think this could become known as conspiratorial machine learning and manipulation.pic.twitter.com/D0pEw73Ywm
Kiitos. Käytämme tätä aikajanasi parantamiseen. KumoaKumoa
-
-
-
The only way I can think of avoiding repeating past mistakes while learning from the data is to change the objective function to have preference over some "correct" data, to still find useful compression patterns in the remaining "bad" data. 1/2
-
Such distinction however still makes bias! An exagerated example. Full dataset: behavior of all humans. "Correct" dataset: behavior of Jesus. The issue of dataset bias will always remain, but it's size might be more limited (growing with the task complexity). 2/2
Keskustelun loppu
Uusi keskustelu -
Lataaminen näyttää kestävän hetken.
Twitter saattaa olla ruuhkautunut tai ongelma on muuten hetkellinen. Yritä uudelleen tai käy Twitterin tilasivulla saadaksesi lisätietoja.