Russian banking giant Sberbank has created a new algorithm, dubbed Auto ML, which serves like a data scientist.
Auto ML is a new model of machine learning, which holds capacity to create other models.
It can solve application tasks, including forecasting client creditworthiness while applying for a loan and assistance to separate law-abiding clients from law-breakers.
In January this year, the first pilot was conducted with Auto ML algorithms where it has been used to create seve baseline models (first versions) of Next BestAction class (targeting of sales campaigns).
Auto ML enables to optimize one of the crucial steps used in business and create baseline model. It is said to be the first version of the model developed with human interaction.
The quality of baseline models created by Auto ML can be compared to the quality of the manually-developed models and algorithm works 10 to 15 times faster than a human, said the bank.
The received results provide flexibility to use automatic modeling technology to quickly form basic models to process data and lacunh sales campaigns for corporate and investment business.
Sberbank executive board deputy chairman Anatoly Popov said: “Introducing AI is one of the possibilities to increase the efficiency of all business processes at the bank. However, the creation of tens and thousands of models to cover all aspects of activity is an almost impossible task if data scientists and developers attempt to create and introduce models manually.
“That’s why we are introducing Auto ML, one of the most advanced approaches in the world to working with machine learning models. The algorithm system that quickly and independently creates application solutions on the basis of machine learning models.”
In November 2018, Russian-based central securities depository National Settlement Depository (NSD), Sberbank and telecommunications operator MTS have completed the commercial bond transaction based on blockchain.
By using blockchain, the NSD, Sberbank and MTS have summed up the results of commercial bond placement.