2nd November 2016

PrivatBank trained neural networks to cope with queues at branches

Ukrainian PrivatBank has implemented a system of service quality analysis based on neural networks. Ukraine's first "smart" system of service quality monitoring automatically tracks data from surveillance cameras installed in every branch of the bank. If there is a queue at one of the bank’s cash offices, the system analyses data and switches the service scenario to "Queue" mode, as well as sends a message to managers to make organisational decisions on reallocation of client flow.

"About a half a million live images from cameras at our branches have been analysed during the process of neural network learning. As a result, the network gained the ability to accurately recognise queues near cash offices at our branches, - said Vadim Kovalyov, Deputy Chairman of the Bank’s Board. - We are working on the application of learning neural networks into the practice of risk identification, which will significantly accelerate and improve the reliability of risk assessment during lending to retail borrowers and entrepreneurs."

Application of a neural network based system for analysing queues will allow the bank to optimise the work of branch staff and to improve customer service. The pilot run of a deep neural network system has shown practically absolute accuracy of queue recognition. On average, more than 50% of all the bank’s branches were recognised by the neural network system as offices without large queues.

According to the developers, in the near future PrivatBank will teach a neural network to recognise ATMs and self-service terminal queues. These data will be used while planning cash replenishment schedules and expansion of the ATM network.

Today, PrivatBank's service network consists of 2,475 branches, 7,246 ATMs and 12,612 self-service terminals.

PrivatBank Press Center

+380 92 716-08-81