Table 1.
Table summarizing the different neural compression schemes used for weak-lensing applications.
| Reference | Loss function | Inference strategy |
|---|---|---|
| Gupta et al. (2018) | MAE | Likelihood-based analysis |
| Fluri et al. (2018) | GNLL | Likelihood-based analysis |
| Fluri et al. (2019) | GNLL | Likelihood-based analysis |
| Ribli et al. (2019) | MAE | Likelihood-based analysis |
| Matilla et al. (2020) | MAE | Likelihood-based analysis |
| Jeffrey et al. (2021) | MSE VMIM |
Likelihood Free Inference (Py-Delfi) |
| Fluri et al. (2021) | IMNN | Likelihood Free Inference (GPABC) |
| Fluri et al. (2022) | IMNN | Likelihood Free Inference (GPABC) |
| Lu et al. (2022) | MSE | Likelihood-based analysis |
| Kacprzak & Fluri (2022) | GNLL | Likelihood-based analysis |
| Lu et al. (2023) | MSE | Likelihood-based analysis |
| Akhmetzhanova et al. (2024) | VICReg | Likelihood Free Inference (SNPE) |
| Sharma et al. (2024) | MSE, MSEPCA, MSENP, VMIM |
Likelihood-based analysis |
| Jeffrey et al. (2024) | MSE | Likelihood Free Inference (Py-Delfi) |
Notes. Gray boxes correspond to analyses performed on real data. Abbreviations used in the Table: MSE-mean squared error; MSENP-mean squared error in S8 space; MSEPCA-mean squared Error in PCA space; MAE-mean absolute error; GNLL- gaussian negative log likelihood; VMIM- variational mutual information maximization; VICReg: variance-invariance-covariance regularization; IMNN- information maximizing neural network; GPABC-gaussian processes approximate bayesian computation.
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.