Since machine learning is relatively new in the financial services space, business leaders must take a fail-forward approach, embracing the notion that not every data initiative will be a knockout success.
Machine Learning is already revolutionizing industries ranging from transportation to healthcare, but, perhaps, no business stands to gain as much as financial services.
This week, I was asked to moderate a panel at Machine Learning X in Toronto. The event brought together a diverse group of professionals across the financial services space, all sharing a common vision for what the future of machine learning might bring to the industry.
Our panel featured speakers from Microsoft, IIROC, Thomson Reuters, Blackrock, and IBM. The panel’s central theme focused on how machine learning can create competitive advantages for the financial services industry. Below are a few of the key highlights from the conversation that ensued.
Machine learning in the enterprise requires making long-term bets
While machine learning holds a great deal of promise for the financial services space, it’s still rather new. As a result, the business impacts and ROI models are equally new, and in many cases, need to be fully proven out. This means that business leaders must take a fail-forward approach, embracing the notion that not every data initiative will be a knockout success.
How is this done? Amir from Thomson Reuters suggested that this is done by making small bets, and validating the results quickly and early. In taking this approach, organizations can quickly understand which aspects of a data initiative are more likely to succeed, and pivot accordingly.
Commit to the cloud
When a member of the audience asked a question about hardware and software resource utilization, the panel quickly agreed, the Cloud (in particular, public cloud offerings such as MS Azure) provide the flexibility for data scientists to scale resources up and down, rapidly and dynamically.
Working in financial services does, however, provide some unique challenges when it comes to dealing with public cloud systems. In particular, privacy and personally identifiable information (PII) need to be considered with the utmost urgency. Hardly a day goes by when there isn’t news of yet another privacy breach or data theft. Keeping customer data and privacy top of mind when designing solutions is absolutely critical for the
Regulators can help, regulators can hinder
One of the most controversial questions that came up during the panel revolved around the role of regulators and whether they strengthen or stifle innovation.
While everyone agreed that there was no way around regulations that exist to protect consumers, not everyone agreed that this needs to slow down the development of machine learning initiatives.
Baiju, Head of Analytics at IIROC, suggested that the restrictions imposed by regulators can actually spur
The panel discussion was punctuated by interesting conversations between business leaders, data scientists, and