101–110 of 177 posts
[flagged]
for those who want some maths-heavy stuff for deep learning, check francois fluret's book https://fleuret.org/francois/lbdl.html . the pdf is free but the print is so cute.
Earlier quoted context omitted.
Bishop’s Pattern Recognition and Machine Learning is one example that has tremendous depth and much clearer notation. Deep Learning by Goodfellow et al. is another example, albeit with less depth than Bishop. I’m glad you’re enjoying the book. The approach is ideal for a very small subset of the ML population, no doubt that was their intention. I’m just weighing in that it’s entirely possible to cover this material w…
Thanks for highlighting Bishop to me! I've self-taught through various resources esp. Goodfellow et al 2016. It's taken me a number of years to rebuild my math knowledge so that I feel comfortable with Goodfellow's treatment and look forward to learning from the Bishop book. Fwiw, I've found the math notation in the Goodfellow textbook to be among the best I've ever seen in terms of consistency and clarity. Some othe…
As someone who has a deeper knowledge of programming rather than math, I find the mathematical notation here to be harder to understand than the code (even in a programming language I do not know). Does anyone with a stronger mathematical background here find it easier to understand the math as written more easily than the source code?
I have a strong mathematical background, and I found the notation completely insane. Right out of the gate in chapter 1 we get a definition that has subscript indices in the subscript index and a summation with subscripts in the superscript, and then composed in a giant function chain. Later we get to 4-level subscripts deep, invent at least 3 new infix operators, define 30 new symbols from 3 different alphabets and…
[flagged]
https://i.imgur.com/DCAnoCu.png
It will also make an attempt at turning the expressions into Python. It bombed out at first but caught itself and retried without any additional prompting:
https://i.imgur.com/nzywKkb.png
Not being familiar with SymPy, though, and not having time to think it through myself, this might be a bunch of hallucinated gobbledygook. Caveat lector.
[flagged]
there are many good materials such as the fantastic fast.ai course that don’t require such mathematical background.
if you are motivated to learn about ML, then studying the topic can gradually be a route in to more mathematical knowledge so that equations like this would not seem intimidating.
[flagged]
This book is just over two months old... Are you saying _you_ did this since Oct 31?
Earlier quoted context omitted.
Use ChatGpt. Screenshot the math, crop it down to the equation, paste into the chat window. It can explain everything about it, what each symbol means, and how it applies to the subject. It’s an amazing accelerator for learning math. There’s no more getting stuck. I think it’s underrated because people hear “LLM’s aren’t good at math”. They are not good at certain kinds of problem solving (yet), but GPT4 is a fantast…
Don't suggest this. While I agree it can be helpful, the problem is if you're a novice you won't be able to distinguish hallucinations. Which in my experience are fairly common, especially as you do advance topice. If you got good math rigor then it's extremely helpful, because often things are hard to exactly search, but it's a potential trap for novices. But if you have no better resource, then I can't blame anyone…
A lot of negativity comes from people who goofed around with 3.X for a while, came away unimpressed, muttered something under their breath about stochastic parrots or Markov chains that sounded profound (at least to them), and never bothered to look any further. 4 is different. 4 is starting to get a bit scary.
The real pedagogical value comes when you try to reconcile what it tells you about the equations with the equations themselves. Ask for clarification when something seems wrong, and there is an excellent chance it will catch its own mistakes.
