Gradient Descent
Neural Networks & Deep LearningChapter 3: Inside the BrainTerm 45 of 253
Optimisation method: take small steps in the direction that reduces error.
At Sunrise School
Gopal asksHow does training reduce the loss?
Bit answersGradient descent: small downhill steps on the error curve. Each step nudges the weights so the next answer is a little better.
In the real worldWalking downhill in fog, one step at a time.
Watch the short
Short coming soonSeason 1 · Day 7Follow @haigopala
Short coming soonSeason 1 · Day 7Follow @haigopala.telugu
Short coming soonSeason 1 · Day 7Follow @haigopala.hindi
Quick check
Quick check. Should every step be as big as possible? Yes or no?
No! Giant steps jump right past the lowest point.
Practise it
Try this term and all 253 others in the free Model Training Arena.
Open the Arena →Follow Bit daily: YouTube English · తెలుగు · हिंदी | Instagram English · తెలుగు · हिंदी