Any business from small companies to big corporations can benefit from machine learning. What are ways of its application? Sergey Vostrikov, Head of Marketplace and Integration at 1C-Bitrix shared his experience of working with machine learning.

Sergey is the author of the book The World of InterBase. Architecture, administration and development of database applications in InterBase. He is also engaged in AI integration projects at Bitrix24.

Speaker: Sergey Vostrikov (S. V.).
Interviewer: AI Conference (A. C.).

A. C.: Is machine learning always related to Big Data?

S. V.: Machine learning is not necessarily related to Big Data. Any kind of data can be used. The point is a level of accuracy. Taking into account the industry case studies, Big Data is not always required.

A. C.: What are the peculiarities of an engineering approach to machine learning?

S. V.: An engineering approach is applied when an immediate solution to a business task is needed. There are plenty of related services and public APIs that provide ready-made black boxes excluding complicated math calculations.

Mentioning an engineering approach, I mean engineering practice that entails the application of theories, technologies and tools in any sphere. Just like programming doesn’t require profound knowledge of math, machine learning is mostly introduced excluding statistical nuances.

A. C.: Could you provide us with a real case study of engineering approach implementation in the sphere of machine learning? What results did you get?

S. V.: Bitrix24 offers a range of services available to end users: call recognition, a CRM face-tracker for identification of clients or employees, working hours tracking, etc. All these scenarios demonstrate the use of ready-made abstract services powered by machine learning that can solve common business tasks.

A. C.: You are Head of Marketplace and Integration at 1C-Bitrix. What does your team do?

S. V.: We develop our platforms of ready-made solutions for Bitrix products, prepare platforms for integration with various external products such as telephony, social networks, technical support services, etc.

This helps us understand practical needs of real clients who want to use Bitrix24 additional software as a principal tool in order to generate maximum revenue. However, do not look for machine learning here. At least, today.

A. C.: Which AI and machine learning solutions does your company develop and use?

S. V.: The list is pretty diversified. It includes the introduction of ready-made solutions with image and speech recognition I have told before and our in-house products. For example, the system of recommendations for online stores and client scoring via a CRM. We practice various learning methods for such a wide range of tasks: from simple linear regression to collaborative filtering.

A. C.: What will you be speaking about at AI Conference in Moscow?

S. V.: My principal objective is to convince people of the practical benefit machine learning gives to a business of any size. We only need to learn how to balance opportunities and limits of machine learning taking into consideration user scenarios.

Say, sometimes a model is accurate to 85%, which is already of benefit to business. However, sometimes even this accuracy is not enough. Math is the same but the results are completely opposite. I will provide real case studies to show how it works.

Sergey Vostrikov will deliver a presentation titled ‘Engineering approach to machine learning. Case studies of victories and defeats’ at AI Conference. To meet the expert and ask him questions, register to the event scheduled for November 22 in Moscow.