Near-term Brexit concerns highlight importance of cognitive computing
Predictions on the effects either result of the United Kingdom’s vote about European Union membership began months before the actual vote. Now that the decision has been made (for the time being anyway) the concern remains.
Most of the talk revolves around the long-term changes, but significant risk is present in the near-term too. Whenever market volatility is present, elements will seek to manipulate trading. Still recovering from the recession, Wall Street traders operating with a short-term focus in a high stakes environment face pressure to capitalize on that volatility. They are being watched by distracted compliance departments struggling to adapt to increased regulatory scrutiny.
Kiran Narsu is the senior vice president of enterprise sales with Digital Reasoning, a cognitive computing software developer. He works with large financial institutions and he sees Brexit’s impact every day.
“We’ve seen increased trading volatility and trading is good for business investment banks.”
Mr. Narsu said he is also seeing a stronger focus on cash control resulting in a more cautious market.
Digital Reasoning helps companies maintain compliance in this new environment, Mr. Narsu explained.
“We can implement and leverage technology to improve compliance and operational efficiency.”
By improving their surveillance companies can indeed reduce risk, Mr. Narsu said. But it is challenging to accomplish, especially if that company is using old technology.
“The problem with regulatory scrutiny is they are looking for collusion and market manipulation which are nuanced behaviors. Past technologies are not good at identifying them.”
The problem is worse because of Brexit which drove many banks’ market caps down, putting pressure on costs and operational efficiencies. The challenge becomes improving performance while keeping costs down and compliance high.
“If you increase surveillance of people with the same models, you’ll only get more of the same data,” Mr. Narsu said.
Digital Reasoning’s cognitive computing technology understands not only language, but also nuances of human communication, he added. Smart organizations consider the impact of macroeconomic events and by leveraging the same cognitive computing technologies they have better opportunities to increase revenue. Machines can be trained on their own to look for behaviors and expand market coverage faster.
“When employing a cognitive approach, all the pieces are taken together in a unified way of reporting knowledge which gives financial institutions a leg up to respond,” Mr. Narsu said. “It makes better use of a massive volume of data.”
Cognitive computing is ideal for guarding against market manipulation because it analyzes all aspects of the trader-client relationship, Mr. Narsu said, a relationship often featuring complex communication.
“There is so much richness in human language data. If someone can accurately extract insights from it it will result in increased revenue, efficiency, and growth.”
Digital Reasoning’s technology makes it easy to onboard and adapt new languages, an important aspect in a globalized marketplace, Mr. Narsu said.
“Multi-lingual deployment — especially to focus in on behaviors of interest across multiple languages — is a challenge for most organizations. Digital Reasoning’s system helps to address this by enabling organizations to analyze behaviors in a given language while understanding the nuance of each conversation. Since our system already understands 12 languages out of the box, it is able to spot conversations of interest, regardless of the underlying language.”
Mr. Narsu used the examples of concealing abuse and collusion, behaviors which are difficult to identify because of the importance of context. Programmers can provide examples of language the system needs to generalize but they struggle to address when an exact phrase should raise concern and when it should not.
The development of algorithms which continually improve as more relevant data enters the system is crucial to the technology’s growth. Because we have so much more data in the digital age, these deep learning algorithms accelerate interpretation.
“We can focus on tangible and measurable benefits, so companies can process more alerts, reduce the volume of noise, and be more productive,” Mr. Narsu said.