You can add location information to your Tweets, such as your city or precise location, from the web and via third-party applications. You always have the option to delete your Tweet location history. Learn more
Machine learning models take the shortest path from input to label, based on previous situations they've encountered -- just like human intuition. Straight input-to-output mapping. And of course, they tend to be highly biased.
Replies indicate that people are very confused about what "bias" means. It means doing pattern recognition based on spurious correlations, as opposed to causal reasoning. A ML model will use all correlations found in the training data, and typically many of them will be spurious.
I would substitute correlations with associations to include non-linear relationships
Well and forensically put.
I have found that the image from the paper https://arxiv.org/abs/1602.04938 of a husky misidentified as a wolf because of snow in the image to be helpful in explaining what you are describing.pic.twitter.com/sMHVyJt1cO
That’s not the statistical definition of bias.
But appreciate the same word is used for a different purpose in ML circles.
One source of confusion might be that sometimes people use "bias" to refer to the properties of a model that *prevent* it from paying too much attention to spurious correlations. E.g.: https://en.wikipedia.org/wiki/Bias%E2%80%93variance_tradeoff …
I would not that bias, personally, but "over-sensitivity to spurious correlation". When the data is gathered as you should in any pure scientific experiment (with proper control), you can mitigate this to a large extent. Right?
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.
