Andriy Radvanskyy Data Scientist, SoftServe
Fields of interest: algorithm research in many fields, including compression, cryptography, error-correcting codes, data mining, optimization.
Fuzzy signatures for email spam filtering
How to process data that you don't have.
How to handle email from all the world on one physical server.
How to rewrite C code in Python and improve performance 100 times.

Tetiana Gladkykh Data Scientist, SoftServe
PhD in Computer Science, Competence Manager - Data Science Group, SoftServe. 
Scientific Interests: data mining, artificial intelligence (genetic algorithms, neural network, and fuzzy logic), mathematical statistic, computer vision, graph theory, game theory and computational mathematics.
Rhythmic and acoustic patterns recognition for musical content recommendation
Presentation focuses on the novel approach for musical content recommendation, which is based on the assessment of musical compositions fuzzy similarity with taking into account their rhythmic and acoustic features. Based on the mel-cepstral analysis, our solution allows to provide recommendations according to the acoustic perception of liked track supported by the results of genre, style and musical direction automatic detection.

Taras Hnot Senior Data Analyst
I have been working at SoftServe Data Science Group since 2015. Since then, I have been specializing in performing data analysis and providing support in making complex business decisions based on it, building predictive models, performing data preprocessing and exploratory data analysis. have an experience in development of anomaly detection systems, analyzing and detecting patterns of huge payment networks, implementing different types of algorithms in order to build computer vision systems, doing NLP analysis etc.
Qualitative content analysis
Now people trying to build bots everywhere to simplify their life. One of the most used area for bots is chats. Bots can communicate with customers, promote products, solve issues or give recommendations regarding them. In one of our projects we were working on development of an AI-driven ChatBot that is able to recognise the topic of the chat conversation to maintain a nearly real-time content feed of relevant articles. Our solution is supposed to help different R&D and engineering teams conduct their research in a more efficient and productive way. In the presentation I will show examples, explain process of implementation, used approaches and data.

Victoriia Shevtsova Developer for Intego Group LLC and Senior Lecture at National Technical University “Kharkiv Polytechnic Institute”
I work Python Developer for Intego Group LLC and Senior Lecture at National Technical University “Kharkiv Polytechnic Institute”. Research interests: machine learning, data analysis, artificial neural network
Methods of topological data analysis for studying the structure of multidimensional data
The Mapper algorithm are method of topological data analysis used for visualization and studying the structure of multidimensional data. This method is based on the idea of partial clustering of the data guided by a set of functions defined on the data. The Mapper algorithm is briefly described and used to analyze the dataset.

Valerii Dmitrienko Dr. Tech. Sci., Professor
National Technical University "Kharkiv Polytechnic Institute" member of Higher Attestation Committee
Neural Networks of Adaptive Resonance Theory: advantages, disadvantages and prospects
The one of the hard challenges in the Neural Network Theory is the problem of stability-plasticity. Various researches in this area have leaded to the development of neural networks of Adaptive-Resonance Theory (ART). However, the practical application of ART neural networks has highlighted their significant disadvantages. The ways of mentioned disadvantages elimination will be considered in the speech, as well as the descriptions of novel ART’s architectures and their applications

Khrystyna Skopyk Computational Linguist at Grammarly
As a computational linguist at Grammarly, I am responsible for developing and improving our error-correction checks. My core competence is natural language processing with Python and Common Lisp.
Breaking the spell of the spelling check
This presentation will overview the task of error correction, with an emphasis on spelling. You will learn how to build a spell checker using the Noisy Channel algorithm. You will see the implementation of this algorithm for the English and Ukrainian languages.

Yuliya Glavcheva Assistant Librarian of Scientific and Technical Library National Technical University «Kharkiv Polytechnic Institute»
Research interests: Infometrics and Scientometrics, Informational technologies in Library; Integration of electronic resources; NLP. Certified trainer of Thomson Reuters. One of the authors and tutors of distance course "Curator of content" (NTU "KhPI").
Formation of the Text Corpus and Identification of Author Style in Academic Works
Qualitative scientific work must be original. Today's information systems are used to detect possible plagiarism to confirm originality. Due to on a lot of possible variants for text presentation, such systems have limited capabilities. But they can be used as additional means of detecting plagiarism. This research shows the features of text corpus creation for the development and testing of software tools for plagiarism detection; experiments with the determination of statistical estimates and lingometry parameters.

Iryna Viblei Data Architect in BI and Big Data CoE
Data Architect in BI and Big Data CoE with major focus on solution design and development of Big Data environments with analytic platforms. Has rich experience in collecting the relevant business information and converting it for the process improvements, enhancing quality of the business and offering architectural solution for the analytical and storage systems.
How Data Lakes provide faster insights
We’ll talk about the power of the data. Try to understand why Data Lakes are an important piece of the overall Big Data strategy and how they transform Big Data into Business Insights. We’ll also talk about the the challenges in building a Data Lake and the key points to think of while designing it.

Igor Kostiuk Computer Networking Professional
Alumni of NTUU “KPI” computer science and philosophical faculty of Taras Shevchenko KNU. Speaker at AI Ukraine, Lviv IT Arena, Kharkov AI Club, Kyiv Deep Learning Meetup, Kyiv Big Data Community etc. Favorite xkcd is 1425, favorite smbc is 1797.Fields of interests: deep learning, reinforcement learning, computer vision.
Deep learning for audio classification
We will discuss methods of audio classification which includes music auto tagging, keyword spotting, urban sound classification etc. Different features extraction approaches and neural network architectures.