Neural networks are helping financial services firms to automate increasingly complex processes and decisions that can ultimately lower costs, improve accuracy and customer experience, and give businesses a competitive edge.
Inspired by the structures of the human brain and initially developed in academia, these artificial neural networks are learning algorithms structured as a number of interconnected layers. They offer a step-change in the power of AI and are especially suited to complex “deep learning” applications that require processing massive amounts of data and high levels of domain expertise and judgment.
Interest in the potential of AI in financial services continues to grow and early proofs of concept for neural networks have yielded promising results. A global banking group now uses facial-recognition technology to analyze over 30,000 reference points to create a map of a user’s face, increasing security and enhancing the customer experience—the company’s mobile app use increased 60 percent in a year.
Neural networks in financial services
- Fraud detection for credit card transactions
- Overdraft predictions based on the customer’s transaction history
- Information retrieval from invoices to substantiate a transaction on business accounts
- Using satellite and street view images to verify the existence of a business as a part of know your customer and anti-money laundering checks
- Predicting health problems and suggesting healthy lifestyle changes by collecting and processing data from wearable devices
- Analyzing customers’ interactions with the company to offer discounts to customers who wish to leave
- Car accident damage assessment
- Image-based risk prediction for home insurance
- Helping traders decide what price to quote when buying or selling bonds for their clients based on historic and real market data
- Extracting information regarding profit or losses from financial reports to aid investment decision-making
- Information extraction and summarization of legal documents
- Automation of site due diligence checks
Some potential risks
Despite the incredible opportunities neural networks present, financial services organizations should be aware of and address potential risks early on. Key concerns are around explainability, algorithmic transparency and bias.
Neural networks were designed to deliver the highest possible accuracy with little focus on explainability, the ability of the algorithm to justify its decisions, and transparency. AI models are largely “black-box” solutions with no native way of articulating the reasoning behind their decisions. The complexity of neural networks’ reasoning also makes explainability challenging.
Another main concern is that it may be difficult to spot bias that could manifest itself through discriminatory outcomes over the long term, leading to unfair treatment of certain groups of customers and potentially legal action for alleged discrimination.
For financial services organizations, the data used for training neural network models should be of sufficient quality, scale and diversity. Organizations should also consider their IT infrastructure, as neural networks require vast amounts of processing power that can only be accessed via the cloud.
Trust and confidence among stakeholders are key to neural networks’ long-term use. Organizations that address risks early on and set up the infrastructure required are expected to unlock competitive capabilities and enhanced value. Download our report where we map out the various components and steps for an effective implementation of neural networks.