Method of and server for training a machine learning algorithm for estimating uncertainty of a sequence of models

A first server for training a first machine learning algorithm (MLA) for estimating an uncertainty parameter of a sequence of computer-implemented models executed by a second server, the sequence of computer-implemented models comprising at least one second MLA, the sequence of computer-implemented models having been trained on a set of training objects to output a set of predictions based on a set of input features, the second server being communicatively coupled to the first server, the first server comprising:

• a processor;
• and
• a non-transitory computer readable storage medium operatively connected to the processor, the non-transitory computer readable storage medium comprising instructions;
• the processor, upon executing the instructions, being configured for:
• receiving a set of labelled digital documents to be processed by the sequence of computer-implemented models;
• receiving, for a given model of the sequence of computer-implemented models, at least one of:
• a respective set of input features,
• a respective set of model-specific features, the respective set of model-specific features comprising parameters of the given model, and
• a respective set of output features predicted by the given model;
• receiving the set of predictions output by the sequence of computer-implemented models;
• and
• training the first MLA based on:
• the set of labelled digital documents, the at least one of the respective set of input features, the respective set of model-specific features, the respective set of output features, and the respective set of predictions output by the sequence of computer-implemented models to estimate an uncertainty parameter of the sequence of computer-implemented models, the uncertainty parameter being indicative of a confidence level of the set of predictions.