poster123

18 ms·

181.
by poster123 9y ago
From the article: "What this means is that, we calibrate classifiers parameters such that it has the same acceptance ratio for all subgroups of sensitive features, e.g. race, sex, etc." But statistics such as crime rates and defau
182.
by poster123 9y ago
The headline is a lie. All students were given more time, in an effort to reduce the ratio of men to women getting the highest grade.
183.
Teamsters Tell UPS: No Drones or Driverless Trucks(wsj.com)
23 points by poster123 9y ago|23 comments
184.
by poster123 9y ago
If someone starts a new company and gets rich, the wealth did not "go" to him -- he created it. On a more prosaic level, my investments in the stock market, which did well last year, represent ownership in companies that employ pe
185.
by poster123 9y ago
It is normal to want grandchildren, and therefore it also normal to be distressed if your children are homosexual.
186.
Deep Learning for Forecasting Stock Returns in the Cross-Section [pdf](arxiv.org)
27 points by poster123 9y ago|2 comments
187.
Of Furries and Fascism at Google(wsj.com)
4 points by poster123 9y ago|0 comments
188.
by poster123 9y ago
The average score of American blacks on IQ tests such as the WISC and Stanford-Binet is about one standard deviation below that of whites, and the performance of blacks on standardized tests such as the SAT and ACT is consistent with that I
189.
by poster123 9y ago
I have wondered why books on machine learning mention cross-validation to avoid overfitting but rarely talk about information criteria such as AIC and BIC. AIC and BIC have been well studied for multiple linear regression models. Are they l