@lachinemearning.com: we all love to throw around the big words backpropagation and gradient descent when praising learning neural nets but it isnt always the answer OMG IGNORE TYPO ON SLIDE 4 it meant to say no gradients needed bye Queen of 0 consistent videos #computerscience #aiml #technology #machinelearning #ai
I mean it’s basically a rule that all dl models end up in a local minimum rather than the global, and even if we could theoretically end up in the global, it would likely yield a model with poor transportability. Momentum and if need be learning scheduling
2026-05-28 19:15:26
0
⃟ :
what on earth does any of this mean
2026-04-19 15:18:43
1
Wander :
a lot of these methods aren’t used since they can’t be optimized or because the prerequisites aren’t satisfied. like newton’s method could be used, but the hessian becomes a problem quickly.
2026-05-06 18:59:12
0
es_modz :
How about the momentum thing
2026-06-05 12:02:01
0
Yonathanan :
Gradient Descent (GD) and Second-Order methods (like Newton’s) are parameter optimizers,they look for the right weights. Evolutionary algorithms and Simulated Annealing are global exploration strategies,they look for where to land. They operate at different levels of abstraction
2026-07-03 02:20:36
0
🕸️🫗💈🗾🎛️📓🔏〽️⚕️ :
You lowkey seem like a chill person
2026-04-19 15:02:13
0
orlo :
Holy song
2026-04-19 16:48:59
1
llamma99 :
I can't wait till I die ☠️, this world is too complex I don't understand anything
2026-04-19 21:59:10
0
To see more videos from user @lachinemearning.com, please go to the Tikwm
homepage.