[Rant] The Male Only History of Deep Learning

all right so instead of reviewing a paper today I thought I might review this thing so this person on Twitter posted this link to an article called brief history of deep learning from 1943 to 2019 of machine learning knowledge dot AI so let's look at this actually

let's look at the tweet first because this is I just saw this the mail only history of deep learning where you say Alex net makes history but imagenet doesn't because women's contributions don't count and contributions from anyone except for white and white adjacent people for that matter that is the tweet and it has 109 retweets over 400 likes and people generally agreeing with this sentiment so the person is expressing concerns that the this article is only going over one particular group of people it so let's

look at the article they basically go over the history of neural networks of deep learning in an algorithmic sense so let's check it out so first we go into neurons starting at 1943 and the perceptron paper right here the first back propagation algorithm from Kelly this actually I think people like schmidhuber would be proud as far as I can tell this is kind of more of a forgotten history or some of these things are more of a forgotten history of course Minsk is paper very famous but here backpropagation attributed to this paper and so on and you can see things people like hint and only coming up later here the Boltzmann machine backpropagation in neural networks now so this as far as I can tell it's just a take on kind of the history of algorithmic development and you can see here it really is about algorithms the algorithms behind deep learning so here is the vanishing gradient problem the LS TM as an architectural component deep belief net deep belief networks then you have GPUs for training again vanishing gradients Alex net and ganz alphago so we're now going a bit faster and then the end it says Turing the Godfather's win the Turing Award for their immense contribution in advancements in area of deep learning and artificial intelligence this is a defining moment for those who had worked relentlessly on neural networks when the entire machine learning community had moved away from it in the nineteen seventies so the article clearly is focused on algorithmic developments in deep learning right and that's why Alex net is here now this person rags that but imagenet isn't and clearly you can see from the article image net is a data set it was not made with deep learning in mind it was simply made as a data set it is not an algorithmic development so ganzar here as well right but the celeb a isn't see for 10 isn't M list isn't the penn treebank isn't right so the I think we've tipped a lot of architectural advancements here like transformers or all kinds of things here but the history is clearly about the algorithmic developments and to reframe this it's clearly States image and it doesn't because women's contributions don't count right the insinuation here absolutely I find this to be absolutely intellectually dishonest and they say and contributions from anyone except for white and white adjacent people that matter at this point you just have to laugh like because of course the narrative that the person wanted to tell was that it's only white people that count but then you scroll in turn it doesn't fit my narrative right this this GPU is not a white person so you to make it fit your narrative you have to call white edge it what is white adjacent it's like if whatever I don't like I now call white and but people just agreeing with this I find this absolutely disgusting and I find the article to be okay I don't know better but if you have a problem with I definitely think there is miss attribution in science throughout even systematic but to say that image net wasn't included because women's contributions don't count that is just a straight out lie and to call people white adjacent is like how does do you not have a bell in your head that goes ding ding ding when you do something like this so I find this to be dishonest either willfully or just because people have so become used to seeing the world in one particular frame and this is I think these calls they only get big whenever there is money and attention going into a field right if you look at like any field where it's just a bunch of weirdos doing their thing the weirdos don't care who's there they just care about the ideas that people have right and I believe we should take that view in science in general I don't care who has the idea and these people do and I disagree all right that was it keep pushing back on these things if you agree as well and keep science for ideas Thanks