Software

Ensemble-deep-learning

GitHub Repository (Jupyter Notebook)

This repository contains a pseudo code for the ensemble deep learning method developed in [1]. The ensemble deep learning method can be used on a classification problem in which the input data is a time series of (also possible multi-channel) images. The ensemble method is based on the optimization of a desired skill score: the implementation comprises the use of standard skill scores as the True skill Statistic (TSS) or the use of the value-weighted skill scores as value-weighted TSS (wTSS) introduced in [2].


[1] S. Guastavino, M. Piana, M. Tizzi, F. Cassola, A. Iengo, D. Sacchetti, E. Solazzo & F. Benvenuto, Prediction of severe thunderstorm events with ensemble deep learning and radar data, (2022) Scientific Reports 12, 20049. 

[2] S. Guastavino, M. Piana and F. Benvenuto, Bad and Good Errors: Value-Weighted Skill Scores in Deep Ensemble Learning, (2022) IEEE Transactions on Neural Networks and Learning Systems, pp. 1-10.


Value-weighted-skill-scores-in-deep-learning

GitHub Repository (Jupyter Notebook)


Forecast verification is a crucial task for assessing the predictive power of prognostic model forecasts and it is usually implemented by checking quality-based skill scores. We introduce a strategy for assessing the severity of forecast errors based on the evidence that, on the one hand, a false alarm just anticipating an occurring event is better than one in the middle of consecutive non-occurring events, and that, on the other hand, a miss of an isolated event has a worse impact than a miss of a single event, which is part of several consecutive occurrences. Relying on this idea, in [1] a notion of value-weighted skill scores is introduced giving greater importance to the value of the prediction rather than to its quality. Then, in [1] two ensemble deep learning methods for binary classification are developed, one based on the verification of standard (quality-based) skill scores and the other based on the proposed value-weighted skill scores.

[1] S. Guastavino, M. Piana and F. Benvenuto, Bad and Good Errors: Value-Weighted Skill Scores in Deep Ensemble Learning, (2022) IEEE Transactions on Neural Networks and Learning Systems, pp. 1-10.

SE-DESAT

A Solar SoftWare tool for image de-saturation in the Atmospheric Image Assembly (AIA) onboard the Solar Dynamics Observatory (SDO)

Software available in SSW-IDL


[1] S. Guastavino, M. Piana, A.M. Massone, R. Schwartz, F. Benvenuto, Desaturating SDO/AIA observations of solar flaring storms, (2019) The Astrophysical Journal Vol. 882:109 (12 pp), Number 2. 

[2] S. Guastavino & F. Benvenuto, A mathematical model for image saturation with an application to the restoration of solar images via adaptive sparse deconvolution, (2021) Inverse Problems, Vol. 37, 015010 (25pp), Number 1.