You mean the entire Apollo program cost $145 billion. This is one mission to the moon with much better technology and lower costs. You just can not compare costs like this.
This is a computer science award, nobody is claiming that AI will revolutionize the fields you mention (which are at best loosely related to CS). It doesn't have to be useful literally everywhere to have value.
The assumption here is that 6) "can't be used broadly" is true. What evidence do you have for this? It turns out this technology is already being used everywhere: text translation, image recognition, autonomous vehicles, text generation and so on. It's already generating huge value which is why so many companies are investing so much in it. There isn't going to be any Winter coming anytime soon.
Do you understand how convolutional neural networks work? There is no segmentation involved here at all. The input are the raw pixels of the image. The output is the probability this image belongs to one of the categories the network is capable of predicting.
Also gradient descent has nothing to do with segmentation at all, I don't understand what you're talking about. Gradient descent is used to find the set of weights that minimizes the error. This is standard in training neural networks of any kind using backpropagation.
No, it doesn't and there is no graph returned whatsoever. It's just a list of the top classification labels for the image (see example at the tutorial he cited https://github.com/tensorflow/tensorflow/tree/master/tensorf...). This is not the result of a segmentation but is rather a list of the top labels the model believes this could be. If you look at the top results you'll see they're usually similar/in the same family (again, refer to the example in the linked tutorial, the top 3 labels are: military uniform, suit, academic gown). This is literally the normalized output of the nodes of the last layer in the neural network (where each node corresponds to one category). If you added all probabilities together it'd sum to 1.
Just a nit, but the author keeps talking about object recognition while what he was actually doing is image classification. Object recognition actually consists of two tasks, one is classifying the object (this is a beer bottle) and the other is also says where in the image the object is. Additionally it can/should detect multiple objects in the image. This is a more complex than classification, which only associates one category with the image.
"There is no data. WhatsApp just passes the encrypted communication from your phone to another phone and that's it. It doesn't hold it."
That's not true. If the user receiving the message is not online how is the message supposed to be delivered? Obviously the server holds it until the user comes online. So yes, there is data.
The title is a little misleading as the context here is art. This is not a general list of deep learning papers to read, but art-related deep learning papers to read. Google's project Magenta objective is to answer the question "can we use machine learning to create compelling art and music?"