Global Veritas Challenge Winners Advance the Cause of Economic Inclusion

Demyst, the provider of a market-leading platform that accelerates the deployment of external data solutions for the world’s leading banks, insurers, and fintechs, and its partner TruEra, a provider of purpose-built Artificial Intelligence (AI) Quality solutions, were named as winners of the inaugural Global Veritas Challenge 2021. The Challenge was organized by the Monetary Authority of Singapore (MAS), the ASEAN Financial Innovation Network (AFIN), and Accenture.

The goal of the Challenge was to spur innovation in AI and data analytics (AIDA) solutions for financial institutions (FIs). The winning project from Demyst and TruEra was recognized for its success in mitigating the effects of bias in credit scoring.

The Demyst platform provided external data to build the project’s machine learning–based credit scoring model. TruEra’s AI solution defined protected groups and set fairness objectives, allowing it to analyze the model for evidence of bias and accuracy issues. Not only did the AI solution identify root causes contributing to the bias, it also conducted targeted interventions.

The collaboration between the two companies demonstrated the power of combining external data with AI quality management solutions to improve the accuracy and fairness of credit decisioning models.

Inequality and discrimination are areas of increasing concern for financial regulators around the world. MAS introduced a set of principles to promote fairness, ethics, accountability and transparency (FEAT) in Singapore’s financial sector. The Veritas initiative lets FIs evaluate AIDA-driven solutions against the FEAT principles.

Top iTechnology Security News: Transmit Security Expands C-Suite, Appoints Chris Kaddaras as Chief Revenue Officer

Scott Albin, General Manager, APAC at Demyst, explained that AI and Machine Learning (ML) models can be made more effective and fair by leveraging third-party data and tools for analyzing and monitoring ML models.

“Machine learning use cases can be both expanded and dramatically improved by the use of high quality, curated, and compliant external data,” said Albin. “This project demonstrates the clear value of external data in improving accuracy and fairness.”