Dense Associative Memories and Deep Learning
ฝัง
- เผยแพร่เมื่อ 12 มี.ค. 2018
- Dense Associative Memories are generalizations of Hopfield nets to higher order (higher than quadratic) interactions between the spins/neurons. I will describe a relationship between these models and neural networks commonly used in deep learning. From the perspective of associative memory, such models deserve attention because they make it possible to store a much larger number of memories, compared to the quadratic case. From the perspective of deep learning, these models make it possible to control the kind of representation that the neural networks learn from a given dataset: small powers of the interaction vertex correspond to feature-based representations, large powers - to prototypes. These Dense Associative Memories can be driven by images processed with convolutional neural networks generally used in image analysis. I will discuss the potential for using this idea to mitigate the problem of adversarial images (very small changes to an input image which lead to a gross misclassification) in computer vision.
See more at www.microsoft.com/en-us/resea... - วิทยาศาสตร์และเทคโนโลยี
Whats the point of focusing on the person instead of focusing on slides..
speaker: ..so we have images like this one *points at smt*
camera: stays dead on speaker
interesting - thanks for sharing... but why are you showing the speaker at times when you ought to show the slides?
Does anyone read these comments? I have commented on many videos about the same problem. Do you realize just how difficult it it to follow the subjects when you throw the slide of for just a couple of seconds and then cut back to the speaker for 99% of the video? I am going to unsubscribe if you don't correct this soon. This is becoming a trend and it's really a shame that "Microsoft" would have such a poor presentation of bleeding edge information. But, such is life. I guess the old corporate saying is true, "people are promoted to their level of incompetence, and there they shall stay."
Bro, do you know how difficult it is to smoothly track a presenter as they move around? It's a skill in itself, one that I appreciate.
I Come here to watch the tracking and don't even have the sound on. My favorite bits are when they do a mad crunch zoom.
Not sure if anybody reads this comments except watchers - or simply there is no way to correct this as the session has already ended.
Probably it is an associative connection between pointer and diagram shown on screen - you need to associate the 3D(X,Y,Z) coordinates when Dmitri's arm move(Left/Right,Up/Down,Forth/Back) and associate it with a location on the image 2D(X,Y).
Anybody else here bc they saw a cat in the clip's thumbnail?
I Know I am. MS Research should put a cat in every thumbnail and the word "cat" in every title.
"Dense Associative Memories and Deep Learning and cat"
this is a prerequisitives must have class... sneaky russian business.