Dr. Kalia please leave these computer science topics to true experts. There is a mistake, misinterpretation, or misunderstanding in your talk almost every minute. You are not qualified to talk about topics in computer science and engineering. Training in neurocritical care, epilepsy, and neurology teaches one nothing about the basics of linear algebra, probability theory, algorithms, or any other basic topic in deep learning. Have you taken these classes? Have you coded a basic MLP in numpy from scratch? 3:20: You say that a single neuron is connected through synapses to the "entire brain". It is well known that synaptic connections even within directly connected layers of the brain are amazingly sparse. A single neuron does not "see the entire brain". You are also very unfamiliar with the recent work of Beniaguev and separately by Aaron Spieler that showd that a single neuron is acutally a complex system where millions to thousands of parameters and hidden units were necessary to model the input-output relationship of spikes. A leaky integrate and fire unit is simply not useful in modeling. 3:29: You have a complete midunderstanding of what "Embarrassingly parallel" means in compuer science as it relates to cognitive science 3:50: You suddenly jump to "transformers" as another catch-word without knowing the basics of transformers and attention. You jump back to a basic fully connected layer (ANN) and say that this is the basis of transformers.
Competition between tech giants is getting better and bigger day by day.
"Mind-blowing! The AMD vs Nvidia GPU war showcased the immense power of GPUs in AI and beyond. Impressive stuff!
Absolutely Amazing, Insightful and Beautifully explained.
Very informative! A much needed detailed video!
This will be a very interesting race between AMD and NVIDIA. Let's see who wins.
Game changing showdown
I've always been a fan of Nvidia. But AMD's technological progress is really interesting!
Dr. Kalia please leave these computer science topics to true experts. There is a mistake, misinterpretation, or misunderstanding in your talk almost every minute. You are not qualified to talk about topics in computer science and engineering. Training in neurocritical care, epilepsy, and neurology teaches one nothing about the basics of linear algebra, probability theory, algorithms, or any other basic topic in deep learning. Have you taken these classes? Have you coded a basic MLP in numpy from scratch?
3:20: You say that a single neuron is connected through synapses to the "entire brain". It is well known that synaptic connections even within directly connected layers of the brain are amazingly sparse. A single neuron does not "see the entire brain".
You are also very unfamiliar with the recent work of Beniaguev and separately by Aaron Spieler that showd that a single neuron is acutally a complex system where millions to thousands of parameters and hidden units were necessary to model the input-output relationship of spikes. A leaky integrate and fire unit is simply not useful in modeling.
3:29: You have a complete midunderstanding of what "Embarrassingly parallel" means in compuer science as it relates to cognitive science
3:50: You suddenly jump to "transformers" as another catch-word without knowing the basics of transformers and attention. You jump back to a basic fully connected layer (ANN) and say that this is the basis of transformers.
Thank you for providing valuable feedback! My content is specifically tailored for non-experts in the field of computer science.