List your company

Image Classification on FMNIST (test) (Accuracy, NLL, ECE)

81.4AccuracyDIMS
12.03230.04148.0566.059May 14, 2026

Evaluation results

MethodLinks
DIMSHessian approximation type=low-rank, Backbone=LeNet, Number of parameter samples=100, friction coefficient η0=0.12026.0581.41.4430.03314.941.4
RLAHessian approximation type=low-rank, Backbone=LeNet, Number of parameter samples=1002026.0576.79.210.0452252
LINLAHessian approximation type=low-rank, Backbone=LeNet, Number of parameter samples=1002026.0538.511.6360.11555.564.7
LAHessian approximation type=low-rank, Backbone=LeNet, Number of parameter samples=1002026.0514.753.5190.16782.586.6