Covid Tackles Bank Machine Learning Models
More than a third of UK banks reported a negative impact on the effectiveness of Machine Learning models following the coronavirus pandemic.
A Bank of England survey of financial institutions found that about 35% of banks reported a "positive" impact of Covid on machine learning (ML) and data science (DS) technologies that support employee teleworking.
New scenarios with the Covid
However, the central bank has found that existing risks may be amplified or that new risks may arise from the use of Machine Learning and Data Science in financial services. For example, ML models may not work well if they are applied to a situation that they have not yet encountered in training data. This is particularly relevant in the context of the coronavirus pandemic, where the underlying data or the statistical properties of the data may have changed.
According to the survey, about 35% of banks also reported a negative impact on the performance of the learning machine model due to Covid. Indeed, the pandemic has caused a severe slowdown that could not have been predicted on the basis of economic data or historical projections alone.
Monitoring to "mitigate risk"
"We will continue to monitor these developments closely with other regulators such as the Financial Conduct Authority and will take the necessary steps to support the safe adoption of ML and SD in financial services.", detailed the Bank of England on its website. "As Covid has led to changes in model performance, more continuous monitoring and validation compared to static validation and test methods is required to mitigate this risk."