From: Fiber laser development enabled by machine learning: review and prospect
Problems | Description | Optimization solution | Learning solution |
---|---|---|---|
Prediction | Solving for y from x when A is unknown | – | finding f: x → y, y = f(x) |
Reconstruction | Solving for x* from specific y* | \(x^{\ast} =\arg \underset{x}{\min }f\left(\mathbf{A}x;y^{\ast} \right)\) | finding f: y → x, x* = f(y*) |
Evaluation | Calculating a evaluation function E(y) with respect to y | – | finding f: y →E(y), E(y) = f(y) |
Design | Finding x = x* to achieve specific y* | \(x^{\ast} =\arg \underset{x}{\min }f\left(\mathbf{A}x;y^{\ast} \right)\) | finding f: y → x, x* = f(y*) |
Control | Solving for Δx to maintain y = y* | \(\varDelta x^{\ast} =\arg \underset{\varDelta x}{\min }f\left(\mathbf{A}\left(x+\varDelta x\right);y^{\ast} \right)\) | finding f: y → Δx, Δx = f(y) |