Projected Gradient Stays Inside

Ball (1.2,0)→(1,0); simplex (0.7,0.5)→(0.6,0.4) by subtracting θ=0.1. Projected GD: w←Π(w−η∇f(w)). Softmax ≠ Euclidean projection.

This Optimization lesson is part of a paid plan. Sign in with a plan that includes this course to read it.