15 days ago
16
0
Posted On: October 21st 2026, 04:03 am
Follow the slope. Reduce the error.
Gradient descent checks the slope at the current position, takes a step downhill, then checks again. The learning rate controls the size of each step. Too large, and it can overshoot. Too small, and progress slows. Different starting points can lead to different valleys.
In daily life: training models to recognize speech, filter spam and recommend music.
Algorithm Series · No. 8.
Explore past algorithms at @ij.muse
#algorithm #gradientdescent #optimization #machinelearning #mathvisualization
Gradient descent checks the slope at the current position, takes a step downhill, then checks again. The learning rate controls the size of each step. Too large, and it can overshoot. Too small, and progress slows. Different starting points can lead to different valleys.
In daily life: training models to recognize speech, filter spam and recommend music.
Algorithm Series · No. 8.
Explore past algorithms at @ij.muse
#algorithm #gradientdescent #optimization #machinelearning #mathvisualization
More posts
