A data scientist uses PROC HPNEURAL in SAS with TRAIN TECH=LEVMAR. What optimization algorithm does LEVMAR specify?
Select an answer to reveal the explanation.
Short Explanation and Infographic
Here's the deal — b is correct because TECH=LEVMAR in PROC HPNEURAL specifies the Levenberg-Marquardt training algorithm, which is a second-order method that interpolates between steepest descent and the Gauss-Newton method. It adapts its step size based on a damping parameter, converging faster than first-order methods for small-to-medium networks.
Full explanation below image
Full Explanation
B is correct because TECH=LEVMAR in PROC HPNEURAL specifies the Levenberg-Marquardt training algorithm, which is a second-order method that interpolates between steepest descent and the Gauss-Newton method. It adapts its step size based on a damping parameter, converging faster than first-order methods for small-to-medium networks. A (SGD with momentum) would be TECH=SGRADIENT. C (Adam) is available in SAS DLPy but not in PROC HPNEURAL's TECH= options. D (L-BFGS) would be TECH=QUANEW in PROC HPNEURAL.