Classification on MNIST (Empirical Risk, Test Risk)
0.065Empirical Risk (Rs(h))ERM
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| ERMModel Type=Linear, Learning Setting=Batch2023.06 | 0.065 | 0.091 | |
| Alg. 1Model Type=Linear, Learning Setting=Batch, epsilon=1/√m, K=0.2√m2023.06 | 0.066 | 0.092 | |
| Alg. 1model=Neural Network, epsilon (ε)=1/√m, K=0.2√m2023.06 | 0.084 | 0.094 | |
| Alg. 1Model=Neural Network, Learning mode=Batch, epsilon=1/sqrt(m)2023.06 | 0.084 | 0.094 | |
| Alg. 1model=Neural Network, epsilon (ε)=1/m, K=0.2√m2023.06 | 0.112 | 0.12 | |
| Alg. 1Model=Neural Network, Learning mode=Batch, epsilon=1/m2023.06 | 0.112 | 0.12 | |
| ERMmodel=Neural Network2023.06 | 0.119 | 0.127 | |
| ERMModel=Neural Network, Learning mode=Batch2023.06 | 0.119 | 0.127 | |
| Alg. 1Model Type=Linear, Learning Setting=Batch, epsilon=1/m, K=0.2√m2023.06 | 0.2 | 0.216 | |
| Alg. 2Model architecture=Linear model, Learning protocol=Online learning2023.06 | 0.284 | 0.31 | |
| OGDModel architecture=Linear model, Learning protocol=Online learning2023.06 | 0.378 | 0.397 |