1D Regression on Synthetic 1D Regression RBF kernel
1.363Context LikelihoodNeural Processes with Stochastic Attention
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Neural Processes with Stochastic Attentionregularization=proposed2022.04 | 1.363 | 0.244 | |
| ConvCNPrepresentation=functional2022.04 | 1.326 | 0.084 | |
| ANPWeight decay lambda=0.0012022.04 | 1.05 | 0.053 | |
| ANPImportance Weighted ELBO samples=52022.04 | 0.895 | -0.031 | |
| ANPregularization=bootstrapping2022.04 | 0.872 | 0.043 | |
| ConvNPrepresentation=functional2022.04 | 0.729 | 0.053 | |
| CNPaggregation=Bayesian2022.04 | 0.575 | 0.112 | |
| CNPImportance Weighted ELBO samples=52022.04 | 0.478 | 0.059 | |
| CNPWeight decay lambda=0.0012022.04 | 0.46 | 0.045 | |
| CNP2022.04 | 0.44 | 0.026 | |
| ANP2022.04 | 0.405 | -0.097 | |
| ANPregularization=dropout2022.04 | 0.372 | -0.163 | |
| NPWeight decay lambda=0.0012022.04 | 0.122 | -0.178 | |
| NP2022.04 | 0.107 | -0.177 | |
| NPImportance Weighted ELBO samples=52022.04 | 0.067 | -0.213 | |
| NPaggregation=Bayesian2022.04 | -0.201 | -0.389 |