One thing to keep in mind is that humans have to process the game from the image on the screen and input through a mouse and keyboard. We have to move the mouse to react to things. The computer is super-human in part because it doesn't have to do these things. It will be interesting to see if they can translate their learnings to bots that react from the image on the screen rather than the API.
I don't usually like to ask for citations, but he said the data was easily found which is clearly false in this case. And there are good reasons to believe google, a company which hires a "VP of Diversity" might be different from the companies previously studied in entirely different industries with entirely different hiring practices and requirements. I know in academia, female applicants are favored in the hard sciences: http://news.cornell.edu/stories/2015/10/new-study-explores-g...
This is a straw-man argument, the point is that if women are predisposed to be less interested in a profession than maybe that explains why there are less woman in that profession. Also the memo is about evaluating everyone based on their merits, not sure how you got the idea that it was for discriminating against women.
There absolutely are people who are against any claims that men and women are biologically different, the president of Harvard got fired for even hinting that this was a possibility. Another way to look at it, if there are differences between men and women than you would expect men to be better at some things and women to be better at some things. Try claiming one of the sexes is better at anything and see how that goes over.
I'd like to see your data supporting the assertion that qualified diverse candidates are rejected at greater rates. The thing is he wouldn't of been fired for arguing the opposite claim, this is clearly discrimination.
Without knowing the distribution of the applicants, we can't say. Maybe 0.1% of all applicants are black, we don't know. In the same way, just saying 1% of tech roles are filled by black engineers doesn't imply that google's hiring practices are discriminatory.
The truth of what he said has nothing to do with why he was fired. He was fired for not being politically correct and offending people, not because his arguments were right or wrong. People are usually most outraged by statements which are true or close to true but which they do not want to believe.
That maybe the theory, but it often isn't how it works in practice. Unfortunately people who claims this publicly get fired. I don't know the numbers, but it would be interesting to see if the acceptance rate of a google job interview depends on gender or race.
Reddit has the worst mobile site i've ever seen, with constant pop-ups that make the site basically unusable. Sometimes a good idea and first to the market is better than good execution.
This is why signing bonuses are so nice, firing someone on day 1 is just awful. If I was an employee at a company who did this to someone I would start looking for another job.
This statement:
"Or in other words: the model, its size, hyperparameters, and the optimiser cannot explain the generalisation performance of state-of-the-art neural networks."
is not true and very misleading. Careful selection of hyperparameters and the model can clearly improve generalization - the article is making a mistake in assuming that getting to zero training error is a good thing or a desirable thing. In fact a large part of hyperparameter optimization are choices that ensure generalization, and some of the fundamental choices such as early stopping and many others do determine how well the model generalizes. If your model has zero training error you have likely made poor choices.
The deep learning buzz-word applies to models which learn feature interactions in a fairly complicated manner. While it is possible to explain what a model is doing, the model is a black-box in the sense that without a computer you could not develop the model. The model is making a lot of decisions, and explaining why all those choices are made is not really feasible. In contrast, I could create a linear model by hand easily, it just won't be as good at complicated predictions. However I think the stigma against black-boxes is undeserved. The human brain is a black-box and no one argues against using it.
You can do the same with deep learning, your model will fail when the train data and test data are systematically different. This isn't about understanding the model, it is about understanding the data. Understanding the data is much more important than interpreting the model. Systematic uncertainty is a problem for all models.
While I have nothing against black-box models and believe that interpretability is over-rated, deep learning is less interpretable than other models. But compare something like a simple linear model with a deep model and the linear model is much easier to interpret, especially for someone without a mathematical background. You don't have to cite papers to give examples of how to interpret a linear model. A single decision tree is also much easier to understand and explain than a deep model. Where deep learning and tree ensemble methods excel is in accuracy and ease of use.