Because I have been studying a little ML and DL in a TinyML course for using DL
for microcontrollers (Simple sensors and activators for IoT kinds of things) I
am starting to read more about DL. I have studied a lot of mathematics but I do
not remember most of it and there are a lot of things that I never studied. So
now that I, at the least, have the beginning of an intuitive clue as to what
the ANN and DL guys are talking about I am starting to be able to pick up on
their writing. For one example, I had no idea why they kept mentioning
matrixes since NNs are not doing matrix multiplication as far as I could tell.
So as I was reading about Gated Recurrent Neural Networks there was a reference
to something called the Hadamard Product. After reading that I could eliminate
one of my guess-work points because the Hadamard (or Shurr) Product is a
slightly different process than what is called the Matrix Product. (I
originally understood the Matrix Product as being something that was derived
from Decartes' 16th century math! Decartes' discriminant has very little (next
to nothing) to do with Matrices in Neural Networks as far as I can tell.) I
sort of figured that must be the case but I had trouble seeing that when they
changed the formulas too often. So there are three problems. They do not give
enough intuitive explanations, they do not give simple examples that you could
potentially follow, and then - if the first two are not enough - they use
established terms differently than you might expect. For another example,
probability in AI is typically quite different than probability in statistics.
AI probability geeks are like a chapter of the Animal House compared to serious
students of scientific statistics. Nothing wrong with that but shouldn't they
mention it? Of course not because everybody who is interested will pick up on
the divergence or they just won't become party animals.
------------------------------------------
Artificial General Intelligence List: AGI
Permalink:
https://agi.topicbox.com/groups/agi/T68be2fedf1f53ef2-Ma2ab6f3e98a9017cc7f86433
Delivery options: https://agi.topicbox.com/groups/agi/subscription
