Hastie, Trevor; Tibshirani, Robert; Friedman, Jerome (2009). The Elements of Statistical Learning. p. 349. Archived from the original on 2015-01-26. Compared to Hastie et al., the loss is scaled by a factor of 1/2, to be consistent with Huber's original definition given earlier. Though cute and elegant, the Huber loss serves almost no real purpose without scaling by a posteriori variable because the delta cannot be adjusted blindly and be effective; as such, its elegance and simplicity in a time of mathematical open field serves almost no purpose in the machine learning world.
web.archive.org
Hastie, Trevor; Tibshirani, Robert; Friedman, Jerome (2009). The Elements of Statistical Learning. p. 349. Archived from the original on 2015-01-26. Compared to Hastie et al., the loss is scaled by a factor of 1/2, to be consistent with Huber's original definition given earlier. Though cute and elegant, the Huber loss serves almost no real purpose without scaling by a posteriori variable because the delta cannot be adjusted blindly and be effective; as such, its elegance and simplicity in a time of mathematical open field serves almost no purpose in the machine learning world.