Open Access

Table 2

Comparison of neural network performance for estimating Mdust and Tdust in different scenarios for 4 different cases.

Mdust (M) Tdust (K)
Case Scenario Bias (dex) RMSE (dex) 3σ outliers (%) Bias RMSE 3σ outliers (%)
S1 (a) −0.0084 0.1696 2.08 1.16 18.16 1.64
1 S2 (b) −0.0786 0.3170 1.72 4.28 31.25 2.05
S3 (c) −0.0135 0.3120 7.01 6.31 59.90 2.02
S1 (a) −0.0110 0.0541 12.80 1.61 14.14 1.45
2 S2 (b) −0.0653 0.1056 15.00 4.79 18.85 2.13
S3 (c) −0.0355 0.1136 16.22 4.60 30.12 1.56
S1 (d) 0.0130 0.2075 8.19 1.39 56.52 1.92
3 S2 (e) 0.0537 0.3985 4.10 −0.62 37.92 2.01
S3 (f) 0.0325 0.5522 2.25 −2.69 78.55 2.34
S1 (d) 0.0013 0.0847 8.78 2.77 42.65 1.41
4 S2 (e) 0.1018 0.1328 21.54 −1.95 20.80 1.72
S3 (f) −0.0235 0.1257 15.72 −8.05 38.52 2.03

Notes. In case-1 and case-2, the training data set contains the preferred subset of JWST filters. In case-3 and case-4, the data set with the minimum subset of JWST filters is used to train our neural network. In case-1, and case-3 the evaluation metrics are applied on the entire test data set. In case-2, and case-4 the evaluation metrics are applied on the predictions of the test data set that have . With the subset of JWST filters that are selected via the feature selection procedure as follows: (a) NIRCam:F070 W, F115W, F140M, F150W, F210M, F300M, F335M, F360M, F430M, F444W, F460M, F466N, and MIRI: F560W, F770W, F1000W, F1130W, F1280W, F1500W, F1800W, F2100W, F2550W (b) NIRCam:F070W, F115W, F140M, F150W, F182M, F187N, F200W, F250M, F277W, F300M, F322W2, F356W, F360M,F405N and MIRI: F560W, F770W, F1000W, F1130W, F1280W, F1500W, F1800W, F2100W (c) NIRCam: F070W, F115W, F140M, F356W, F460M, F480M, and MIRI: F560W, F770W, F1000W, F1130W, F1280W, F1500W, F1800W (d) NIRCam: F460M, and MIRI: F560W, F770W, F1130W, F1280W, F1500W, F2100W (e) NIRCam: F140M, F150W, F200W, F300M, F356W, F360M, F410M, and MIRI: F560W, F770W, F1000W, F1280W, F1500W, F1800W (f) NIRCam: F070W, F140M, F356W, F480M, and MIRI: F560W, F770W, F1000W, F1130W, F1500W, F1800W.

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.