The research associate in IMMSP NASU. More than 10 years of experience with machine learning algorithms, neural network algorithms, image recognition algorithms.
Headed the direction of Computer Vision in ZZ Wolf, previously participated in development of intelligent algorithms for a number of organizations, including Neurologix Security Inc., Postindustria, Samsung, US Air Forces.
Reviewer of the scientific journal "Neural Networks" (Elsevier).
How to confuse a customer head in Data Science-projects
11:40—12:20. Internet technologies
In our days is rapidly grows demand for "smart" algorithms from pool of Data Science, for the solution of such tasks as: images recognition, determining of text tone, targeting of advertising, recommendation systems and more.
At the same time, for such algorithms by their nature are almost never guaranteed absolute accuracy of work. The estimation of work quality can be from 0% to 100% correct answers, depending on the test method, used data and desire of developer to show good work of "smart" algorithm.
The author addresses the report to customers, that wishing to improve the quality of project evaluation, connected with Data Science. He provides standard test methods of quality of such systems, discusses the stumbling blocks and various developer tricks for creation of good impression about quality of project.
Sponsors 2016
Information Partners 2016
Partner
Partner
Information Partner
Partner
