The leading knowledge platform for the financial technology industry
The leading knowledge platform for the financial technology industry

A-Team Insight Blogs

SmartStream Offers Affinity Machine Learning Technology in the Cloud

SmartStream Technologies has acknowledged banks’ ongoing move to cloud services and growing interest in machine learning (ML) for data management and discovery with its offer of Affinity, an observational learning solution designed to improve operational data management and data quality processes.

Affinity observes how users work, learns from them, and incrementally increases its understanding of the workflow over time. Once it is trusted, it can run in batch mode with users handling only exceptions. The initial use case of the technology is reconciliations. One Tier 1 bank collaborating on development with SmartStream, is expected to achieve cost savings of up to $2 million a year.

SmartStream CIO Andreas Burner, says: “Affinity can boost matching beyond typical levels of around 90%, provide cost savings, and help to optimise the business. How much better it can get at matching depends on the maturity of the bank and the quality of data used to train the ML model.” Like most ML solutions, Affinity can address both structured and unstructured data to identify relationships beyond those that can be discovered by human resources.

Affinity is a product of the SmartStream Innovation Lab and has been in the making for about a year, and has been successfully tried and tested by seven banks – you can read about SmartStream’s approach to AI to improve reconciliations here.

It is based on Python standard libraries, although Burner notes that while python is the language of choice for ML, it is difficult to scale, leading SmartStream to start testing Google Code’s new RAFT language that includes parallelisation to provide the computing power required by ML.

The solution is available from today in SmartStream Air, the company’s cloud native AI platform, and provides a multi-tenant service with encryption to hold down costs. It will also be embedded in the company’s key reconciliation solutions TLM Aurora and TLM Reconciliations Premium. Importantly, it will also be the basis of other ML solutions spanning financial services and taking SmartStream into other sectors including healthcare and insurance. Or as Burner puts it: “Affinity is about comparing data – that could be anything in the world in any sector.”

He says the company will continue to work with customers to improve Affinity technology for reconciliations over the next few months before embarking on additional use cases, perhaps in human resources, where a lot of data must be matched, and data migration, where banks move to new software and want to ensure no data is lost.

Related content

WEBINAR

Recorded Webinar: Improving data integrity to address regulatory requirements

Financial institutions today face a global regulatory landscape characterised by rigorous and varied reporting requirements across their businesses. Reporting challenges include completing more data fields across more lines of business with greater frequency, adding complexity and cost. At the same time, there is waning tolerance among supervisory bodies for errors, issues or delays – as...

BLOG

Green Data – Searching for ESG Data and Helping Define ESG Data Standards

By Janine Hofer-Wittwer, senior product manager, SIX Financial Information. Global adoption of ESG strategies by investors and corporations has risen astronomically in recent years. The overall value of assets under management at funds leveraging ESG data has increased significantly, from US$22.9 trillion in 2016 to US$40.5 trillion in 2020, according to data from Opimas. It...

EVENT

RegTech Summit APAC Virtual

RegTech Summit APAC will explore the current regulatory environment in Asia Pacific, the impact of COVID on the RegTech industry and the extent to which the pandemic has acted a catalyst for RegTech adoption in financial markets.

GUIDE

Entity Data Management Handbook – Seventh Edition

Sourcing entity data and ensuring efficient and effective entity data management is a challenge for many financial institutions as volumes of data rise, more regulations require entity data in reporting, and the fight again financial crime is escalated by bad actors using increasingly sophisticated techniques to attack processes and systems. That said, based on best...