One should realize machine learning opportunities: Aleksandr Serbul, 1С-Bitrix
The efficiency of machine learning in electronic commerce depends on how clearly you realize its opportunities. It is the opinion of Aleksandr Serbul, a speaker at the Internet of Things conference and the Head of Integration and Implementation Quality Control Department at 1С-Bitrix. We talked to him about machine learning algorithms applied in work processes and technology advantages in e-commerce.
Interviewer: IoT Conference (IoT).
Respondent: Aleksandr Serbul (AS).
IoT: Hello, Aleksandr. Please tell us what machine learning algorithms you apply in your work.
AS: We use a wide spectrum of algorithms: from statistical to convolutional as well as recurrent and competitive neural networks. Initially, we try to choose the simplest and properly operating solutions rather than get involved in the maze. Our team actively works with the data, applying various unsupervised teaching methods: it helps to be well on the way to the selection of algorithms.
IoT: How popular are e-commercial solutions based on machine learning on the market?
AS: Their demand constantly grows, as this area allows to increase conversion and profitability in businesses. Our platform offers several thousands of online stores; all of them are based on our machine learning and Big Data cloud services.
IoT: What are the advantages of machine learning in e-commerce?
AS: There is a possibility to overcome competitors using predictive analysis methods: they allow to take a look into the future. Besides, according to our experience, project conversion grows and expenses decrease. It is caused by the fact that tasks previously requiring certain specialists are now automated.
IoT: What project in your practice was the most complicated in engineering terms?
AS: The most complicated was the project of neural chatbot suggesting the answers to the questions. We developed it along with the Moscow City Government. A recurrent combination of neural networks was poorly converged, and it was a major challenge to find a training set in Russian. But we coped with this issue.
IoT: Tell us in detail about your presentation at the Internet of Things conference.
AS: The presentation will explain in layman's terms the concepts of efficient application of machine learning in e-commerce in the context of the Internet of Things: from voice assistants and columns to smart homes and pizza order without leaving your bed. We will just share our vast experience in this sector and on the Russian market.
The future is coming faster than we expect. Therefore, the key to success is in the clear intuitive understanding of machine learning capacity in business tasks.