Efficient training of interval neural networks for imprecise training data

Sadeghi, Jonathan and de Angelis, M. and Patelli, Edoardo (2019) Efficient training of interval neural networks for imprecise training data. Neural Networks, 118. pp. 338-351. ISSN 0893-6080

[img] Text (Sadeghi-etal-NN-2019-Efficient-training-of-interval-neural-networks-for-imprecise)
Sadeghi_etal_NN_2019_Efficient_training_of_interval_neural_networks_for_imprecise.pdf
Accepted Author Manuscript
Restricted to Repository staff only until 29 July 2020.
License: Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 logo
Download (721kB) | Request a copy from the Strathclyde author

    Abstract

    This paper describes a robust and computationally feasible method to train and quantify the uncertainty of Neural Networks. Specifically, we propose a back propagation algorithm for Neural Networks with interval predictions. In order to maintain numerical stability we propose minimising the maximum of the batch of errors at each step. Our approach can accommodate incertitude in the training data, and therefore adversarial examples from a commonly used attack model can be trivially accounted for. We present results on a test function example, and a more realistic engineering test case. The reliability of the predictions of these networks is guaranteed by the non-convex Scenario approach to chance constrained optimisation, which takes place following training, and is hence robust to the performance of the optimiser. A key result is that, by using minibatches of size M, the complexity of the proposed approach scales as O(M⋅Niter), and does not depend upon the number of training data points as with other Interval Predictor Model methods. In addition, troublesome penalty function methods are avoided. To the authors’ knowledge this contribution presents the first computationally feasible approach for dealing with convex set based epistemic uncertainty in huge datasets.

    Creators(s): Sadeghi, Jonathan, de Angelis, M. and Patelli, Edoardo ORCID logoORCID: https://orcid.org/0000-0002-5007-7247;
    Item type:Article
    ID code:71230
    Keywords:imprecise probability, interval predictor models, machine learning, neural networks, uncertainty quantification, Risk Management, Cognitive Neuroscience, Artificial Intelligence
    Subjects:Social Sciences > Industries. Land use. Labor > Risk Management
    Department:Faculty of Engineering > Civil and Environmental Engineering
    Depositing user: Pure Administrator
    Date deposited:28 Jan 2020 12:03
    Last modified:29 May 2020 02:59
    Related URLs:
    URI:https://strathprints.strath.ac.uk/id/eprint/71230
    Export data:
    CORE (COnnecting REpositories)