DATA SCIENCE UA

CONFERENCE

KYIV
MARCH 16
2019
×
  • THE KNOWLEDGE IS EVERYTHING

    Data Science UA Conference brings together leaders in machine learning,
    analytics, BI, data science, AI for a day-long exploration of
    how data trans-forms the World today and what is going to be tomorrow.
    On March 16 we will be gathering for the 6th time in Kyiv.

    400
    Participants
    22
    Speakers
    3
    Tracks
    10
    Hours of networking

    Stages


    Tech

    Business

    Workshops

    Panel discussion

    Speakers

    PANEL DISCUSSION

    Opportunities and Challenges for Artificial Intelligence: from theory to real life

    Hosts

    Daniel Che

    Comandante

    Che – Guerrilla Marketing

    Jane klepa

    Executive director

    1991 Open Data Incubator

    Tickets

    WE MAKE A TEAM TOGETHER

    BUY 2 TICKETS FOR EARLY BIRDS PRICE
    ONLY ON FEBRUARY 14
    Buy 5 tickets — Get 10% off
    Buy 10 tickets — Get 15% off
    25% discount for students
    1450
    January 15 — January 31
    2150
    February 1 — February 28
    2750
    March 1 — March 15
    3000
    March 16

    Become one of our Partners

    The Data Science UA Conference attracts both, the best and brightest attendees, and many of the top companies using data science.

    Partners

    General partner

    Silver partners

    Bronze partners

    General HR partner

    Book partner

    Media Accreditation

    APPLY NOW >

    Become a volunteer

    APPLY NOW >
    Location

    Oasis concert hall Kyiv

    Kyiv, Lipkovskogo str, 1A
    Ultramarine, 3rd floor


    Contact information

    +38 099 055 23 92
    info@data-science.com.ua
    #datascienceua


    ×

    Andrii Burlutskyi

    Strategic marketing manager, SMART business

    Keynote Language: Russian

    Intelligent Economies: AI’s Transformation of Industries and Society. The promise of AI holistic transformation.

    Digital Transformation Expert and Fan of Our Future, having 14 years' experience in IT business, 10 of which at Microsoft. Keynote speaker at leading conferences: Dynamics Day, SMART talks, iForum, Forum of Marketing Directors, Ideas Days in KMBS, DeForum Microsoft and others.

    His purpose is to build trust to technologies and storytelling

    Prof. Marcel Worring

    Director of the Informatics Institute and Full Professor Amsterdam Business School, University of Amsterdam

    Technical track Language: English

    (Hyper) graph convolutional networks for analyzing social multimedia data.

    Prof. dr. Marcel Worring is a full professor in Data Science for Business Analytics at the Amsterdam Business School of the University of Amsterdam (UvA). He is the director of the Informatics Institute of the UvA where he is an associate professor and does most of his research. He was an associate director of Amsterdam Data Science a network based organization with companies and researchers to promote data science in Amsterdam. He is also one of the co-founders of the recently launched Innovation Center for Artificial Intelligence in which companies / government organizations and knowledge institutes work together on a common research agenda in labs of at least five sponsored PhD students.

    We will give an overview of different methods of face transfer and our own system based on conditional generative adversarial networks. Also, we will talk about the problems we met with temporal video frame consistency and the ways to solve it.

    Oles Petriv

    CTO, NeoCortext / ML Lead, Videogorillas

    Technical track Language: ukrainian

    Face transfer on video using conditional GANs

    Oles has been involved in active research and development of computer vision and natural language processing systems for the last 6 years. He is the author of Machine Learning course on Prometheus and Deep Learning course in ARVI Lab. Oles has a strong background in video processing using deep learning methods for object and action detection, image and video captioning for Hollywood movie studios

    We will give an overview of different methods of face transfer and our own system based on conditional generative adversarial networks. Also, we will talk about the problems we met with temporal video frame consistency and the ways to solve it.

    Nazar Shmatko

    VP of engineering, NeoCortext

    Technical track Language: ukrainian

    Face transfer on video using conditional GANs

    Nazar, as a specialist in machine learning, has experience in computer vision, natural language processing, and healthcare. He is currently actively working with generative models for the project to transfer facial rice.

    We will give an overview of different methods of face transfer and our own system based on conditional generative adversarial networks. Also, we will talk about the problems we met with temporal video frame consistency and the ways to solve it.

    Elizaveta Lebedeva

    Data Scientist, Taxify

    Technical track Language: russian

    Experiments in Data Science: beyond AB-tests and p-values

    Elizaveta Lebedeva works as Data Scientist at Taxify. Her main focus is supporting lifecycle marketing campaigns, ensuring the company growth and delivering the best experience for riders and drivers. Being passionate about math and having degrees in finance and economics, she transitioned to Data Science from Business Analytics, marking her path with numerous math competitions and hackathons.

    Experiments are a key part in applied data science. Running A/B tests (either for models or for product features) and evaluating results with t-tests is common practice. But there are much more methods to perform experiments. This presentation will cover causal inference, multi-armed bandit (MAB)-based continuous experiments, as well as frequent pitfalls in A/B/N tests and guidelines on how to avoid them. In addition, I talk about different types of metrics used in experiments and how to improve test evaluation with outlier detection, variance reduction, and pre-experiment bias.

    Anton Vokrug

    CEO, DataTrading — AI for smart trading decisions

    Business track Language: Russian

    Taming Neural Networks or In Search of the Soul of Artificial Intelligence

    He have over 10 years of experience in IT entrepreneurship. Successful sale of several businesses. Marketing and Sales in IT. Blockchain and ICO support.

    Real experience in developing commercial products based on artificial intelligence algorithms. Expectations and reality. Curiosities, fears, and joys of the achieved results.

    Eugene Potemskyi

    COO, DataTrading — AI for smart trading decisions

    Business track Language: Russian

    Taming Neural Networks or In Search of the Soul of Artificial Intelligence

    Eugene is a COO at DataTrading (which has offices in China and the United States), which deals with the prediction of financial markets for artificial intelligence. Eugene constantly speaks at international conferences devoted to blockchain and artificial intelligence. Basic skills and competencies: investment analysis (more than 10 years), development and management of systems for working with large data (more than 4 years).

    Real experience in developing commercial products based on artificial intelligence algorithms. Expectations and reality. Curiosities, fears, and joys of the achieved results.

    Ruslan Gutnikov

    Executive Director, OMD MD Intelligence

    Business track Language: Russian

    Brand Metrics®: How Data Science helps boost the ROI of the bank's advertising campaign

    Ruslan is an ambassador of introduction of data-based marketing solutions. He has 8 years of experience in the country's largest advertising holdings, which resulted in expertise in the field of media and communication planning, a deep understanding of market opportunities and business needs. Today, he is responsible for the development of the OMD MD Intelligence analytical unit. The main focus of the team is the introduction of data science technologies to optimize marketing investments, provide integrated marketing analytics, strategic customer support.

    Bank calls to the Call Center, online application forms and visits to the branch offices themselves are the main channels for the sale of consumer products. The main purpose of any advertising campaign is to attract attention to the bank's offers and increase the number of customer queries. Thanks to our Brand Metrics® approach, we can clearly identify and distinguish the impact factors on the feedback of potential customers, determine which communication channels to use to increase inflow of queries. In the speech, we will talk about one component of this project - a case for forecasting and managing calls in the Call Center. We will talk about how to increase the effectiveness of advertising campaign by 8% in the first 2 months. We will go through the key stages of implementing our approach, directly in the process of machine learning and dwell on the important technical aspects of the implementation of forecasting models.

    Yana Fareniuk

    Head of Data Science Analytics, OMD MD Intelligence

    Business track Language: Russian

    Brand Metrics®: How Data Science helps boost the ROI of the bank's advertising campaign

    Yana is a leading specialist in the development and implementation of mathematical modeling projects and machine learning in the media field. Over the past 3 years, she has managed to implement more than 50 optimization projects for clients in the financial category, telecom, pharmaceutical industry, e-commerce, online and offline retail. Yana Fareniuk has a Master Degree specializing in "Economic Cybernetics", Kyiv Taras Shevchenko National University. Today, she is responsible for customer support through the Brand Metrics® approach, developed by OMD MD Intelligence, based on machine learning technology and aimed at optimizing customer marketing campaigns.

    Bank calls to the Call Center, online application forms and visits to the branch offices themselves are the main channels for the sale of consumer products. The main purpose of any advertising campaign is to attract attention to the bank's offers and increase the number of customer queries. Thanks to our Brand Metrics® approach, we can clearly identify and distinguish the impact factors on the feedback of potential customers, determine which communication channels to use to increase inflow of queries. In the speech, we will talk about one component of this project - a case for forecasting and managing calls in the Call Center. We will talk about how to increase the effectiveness of advertising campaign by 8% in the first 2 months. We will go through the key stages of implementing our approach, directly in the process of machine learning and dwell on the important technical aspects of the implementation of forecasting models.

    Tetiana Hladkykh

    Research Data Scientist, SoftServe
    PhD in Technical Science

    Business track Language: Russian

    Predictive Analytics in Human Resources

    Tetiana is a Senior Data Scientist with strong computer science background. Her area of scientific interest includes Data mining, Artificial Intelligence, Mathematical Statistic, Computer Vision, Graph Theory, Game Theory and Computational mathematics. Tetiana has 18 years of academic experience and is an author of more than 35 scientific publications in fields of Electronic circuit simulation and Applied problems of Artificial Intelligence. Additionally, Tatiana was a research advisor in 26 scientific projects and publications in Computer Vision, Multi-criterion optimization and Economic planning and forecasting.

    Human Resource analytics is addressed to the analysis of any employees-related problems in any organization. The most significant problems facing by managers are connected with employee turnover. Turnover numbers were always important for any organization, especially in the case of their continued growth – the more the organization is growing the more these numbers become critical. We are presenting our solution in the context of HR analytics aimed to notify direct people managers about the risks of their employee leaving. Our data-driven solution with the state-of-art engine that uses machine learning approaches to help managers proactively manages unwanted dismissals.

    Dmytro Zikrach

    Data Scientist, SoftServe
    PhD in Mathematics

    Business track Language: Russian

    Predictive Analytics in Human Resources

    Dmytro successfully combines knowledge in Mathematics and practical approaches in the Data Science area. He has two Master's degree (in Mathematics, and Finance), and he has the Ph.D. Dmytro has more than 50 scientific publications. His scientific interest lies in the area of Mathematical and Complex Analysis, Theory of Probabilities, Machine Learning and Predictive Analytics, Computer Vision and Time-series Analytics. Dmytro has great experience and research in Statistical Learning, Predictive Analytics, NLP, Time Series Analysis, Artificial Intelligence and Recommender Systems. Moreover, he often takes part in various Data Science community competitions, hackathons, and challenges.

    Human Resource analytics is addressed to the analysis of any employees related problems in any organization. The most significant problems facing by managers are connected with employee turnover. Turnover numbers were always important for any organization, especially in the case of their continued growth – the more the organization is growing the more these numbers become critical. We are presenting our solution in the context of HR analytics aimed to notify direct people managers about the risks of their employee leaving. Our data-driven solution with the state-of-art engine that uses machine learning approaches to help managers proactively manages unwanted dismissals.

    Ksenia Demskaya

    Research Engineer, Ciklum

    Technical track Language: ukrainian

    Adversarial attacks on deep neural networks

    Ksenia had graduated from Institute of Physics and Technology of the Igor Sikorsky KPI with a degree in Applied Physics. Currently she's working as a research engineer at Ciklum in the area of Deep Learning applied to computer vision tasks, such as image classification, object detection and segmentation. Her research interests cover the theoretical aspects of Deep Learning and problems of interpretability of convolutional neural networks.

    Recent advances in Deep Learning show the existence of well-designed input samples, called adversarial examples, that can fool a state-of-the-art network and cause it to make a mistake. We will discuss the following questions: 1) what are the adversarial examples 2) how to generate the adversarial attacks and to defend against them 3) what are the current challenges in this field of research 4) do we need to be aware of adversarial attacks

    Volodymyr Budzan

    Research Engineer, SoftServe R&D

    Technical track Language: ukrainian

    GANs for geophysics applications

    Volodymyr works as a Research Engineer at SoftServe R&D department. Mostly interested in Deep Learning and Computer Vision, he combines working on commercial projects and doing scientific research. He is a co-author of scientific publications presented at CVPR 2018 and WiML 2018. Currently Volodymyr is doing a research in seismic images enhancement with GANs

    It takes experts several years to collect and interpret seismic data to analyze subsurface layers for various applications. This duration can be reduced to several months with the progress in deep learning for computer vision applications. In this talk I will go over the nature of seismic data, the problems that arise during its interpretation and how we could approach this data with deep learning and generative adversarial networks (GANs).

    Alexandr Honchar

    AI Solution architect, MAWI

    Technical track Language: ukrainian

    GAN vs ODE: the end of mathematical modeling?

    Alexandr is a machine learning expert with experience in computer vision and time series analysis, who worked with Ukrainian (MAWI, ARVI), Russian (Mlvch), American (Inma AI), and Italian (HPA Srl) companies. He is mainly interested in bringing research ideas to production and making value from latest theoretical developments in AI. Currently he is working in MAWI (USA-based company with R&D department in Ukraine) on biomedical signal analysis, in particular ECG, and applying machine learning for classical applications like medical diagnostics and developing novel cases as well. Meanwhile, Alexandr teaches deep learning seminars in University of Verona and writes a popular blog on Medium.

    For hundreds of years, scientists were developing strong theories and rigorous mathematical models to explain patterns and dependencies in data and processes around us. Today instead of modeling some features of data by ourselves we rely on deep neural networks and they don't let us down. So, the natural question arises: do we really need human experts to describe the world mathematically or let's just let AI do all the work? In this talk, we will connect dots between generative neural networks (GANs) and mathematical models like ODEs (ordinary differential equations) for ECG analysis: a classical area driven by pure mathematical models for decades. We analyze empirically what human experts did and what neural networks have learned by themselves and will try to understand, how close we are to fully rely on AI in ECG analysis and other areas.

    Nadiia Romanenko

    Data journalist, Texty.org.ua

    Language: ukrainian

    How Texty.org.ua applied transfer learning to detect manipulative news

    Texty.org.ua has recently released longread “We’ve got bad news”. The project is based on language model classifier. It detects manipulation in news’ texts so that we can compose a ranking of toxic news outlets. We will talk about technical details: the choice architecture, key challenges and solutions. Our case is both about classification of news and application of deep learning in small team.

    Volodymyr Kornienko

    Head of the loyalty program of the Ukrainian retail chain "Chervonyi Market"

    Business track Language: Ukrainian

    AI infusion into Marketing: A Personalized or Personified Marketing Campaign. How to create trouble-free offers for the customer? Story of increasing the conversion in retail network Chervoniy Market.

    5 years of experience in retail, of which 3 - development of loyalty program and marketing. Admire the technology and psychology of human behaviour. He sure, that the infusion of innovations in the most economy segment - way to develop both business and consumers in Ukraine. Create an effective system of personalized customer interaction, that benefits consumers and boosts business profits.

    Dmitriy Solopov

    Business Development Manager Data, AI and Advanced Analytics, SMART business

    Business track Language: Russian

    AI infusion into Marketing: A Personalized or Personified Marketing Campaign. How to create trouble-free offers for the customer? Story of increasing the conversion in retail network Chervoniy Market.

    His mission is to serve SMART business customer, using innovations in analytical and cognitive services to increase business effectiveness. Dmitriy spent 7 years developing technology and strategy at Microsoft. And 4 years as certified trainer experienced in readiness programs development, official Microsoft courses. Guest speaker at leading conferences: Predictive Analytics World (DE), Dynamics Day, DeForum, SMART talks, Idea Days at KMBS and other.

    Halyna Oliinyk

    Natural language processing engineer, 1touch.io

    WORKSHOP Language: Russian

    Named entity recognition in the wild: intro to development of tensorflow-based system as a component of data lifecycle management solution

    Galina is a former natural language processing engineer of 1touch.io, platform for advanced data lifecycle management. She has long-term experience of continuous delivery of end-to-end NLP solutions mainly focused on performing multi-lingual analysis for high-load systems helping to enhance, summarize and highlight specific units/properties of the text data. Her field of R&D interests also includes handling database processing, running big data analysis on the large scale and implementing existing NLP workflows described in scientific papers. Today, she is responsible for the development and support of solution-specific NLP module of the product, which is based on both classical natural language processing algorithms and deep learning pipelines.

    A workshop consists of two parts: an introductory speech on the development of state-of-art named entity recognition model, which includes recurrent neural network and sequence-to-sequence conditional random fields layer, and a workshop itself, which is dedicated to the development of end-to-end tensorflow named entity recognition neural network. The presentation part includes an introduction into training data specifications for DLM product, advances of feature engineering and word embeddings usage/combination, framework-specific peculiarities of appropriate model design, and Google Cloud ML Engine capabilities for running the solution using GPU/TPU. Workshop part can be divided into 5 main steps: 1. defining and explaining tensorflow data types to initialize future model placeholders, creating data encoders and preprocessors; 2. constructing tensorflow session, talking about the graph and its properties; 3. building basic neural network architecture, explaining parameters and cell types; 4. running the session with previously created feed dict for a few epochs, retrieving and interpreting the training loss; 5. calculating final precision and recall. The needed tools are python >= 3.4 (ideally with previously created pyenv virtualenv) and tensorflow >= 1.12.0. All of the additional packages are light-weight and can be easily installed even with slow wi-fi. The link to the base repo for the workshop will be shared on the event. Recommended reading is https://www.tensorflow.org/tutorials/ (but not required).

    Nika Tamaio Flores

    Head of Consulting, Data Science UA

    Moderator of Panel Discussion

    While working in business, I was faced with tasks that were difficult to solve by the well-known methods or these methods did not give the desired solutions. Interest in technology and the search for answers led me to data science. I studied at IE Business School in Business Analytics & Big Data program. By combining new knowledge gained during my studies with business experience, I am now helping companies to benefit from the data they collect and make data-driven decisions.

    Oleg Boguslavskyi

    General Manager, Ring Ukraine

    Panel Discussion Language: Russian

    Graduated from Faculty of Applied Mathematics and Faculty of Management (Second Higher Education) at NTUU “KPI” in 2003-2004. Worked in embedded software development for SoC and MPSoC for more than 16 years for the companies Motorola/Freescale, Mindspeed, AMD, LGE, NVidia, Renesas. Holds leading management positions for 12+ years. Has broad experience in running remote teams in USA, China, France, Russia, etc. Currently runs the organization of ±150 highly-qualified specialists. Has 9 scientific publications and technical expertise in: -VoIP; -Video codecs and real-time video streaming -GPU compute (including neural networks optimization for embedded GPU) -Building fault-tolerant real-time systems -Augmented navigation in Automotive industry.

    Dimitri Podoliev

    Founder, iHUB

    Panel Discussion Language: Russian

    Не got an engineering degree from Cambridge University and MIT. Over the past decade, he founded one of the largest tech accelerators in Eastern Europe as well as was on the core team of an Agriculture Company which grew from zero to 2000+ employees with over $300mln in total invested capital. His specialties are hybrid mobile app development, deep neural networks, machine learning, smart contracts and blockchain-based tools and niche mobiles marketplaces.

    Bohdan Pavlyshenko

    Data Scientist (Ph.D.), SoftServe

    Panel Discussion Language: Russian

    Bohdan combines academic theories and practical approaches in the data science area. He is a Data Scientist at SoftServe and a postdoctoral researcher at Electronics and Computer Technologies Faculty at Lviv National University. He has more than 50 scientific publications. His current scientific interest lies in the area of quantitative linguistics, machine learning, predictive analytics, computer vision, social network mining, business intelligence, time series analytics, numeric modeling, risk assessment, reliability theory, financial modeling. He has practical experience in retail and supply chain analytics, customer’s behavior analytics, fraud detection.

    He works on the state-of-the-art predictive analytics solutions, taking part in Kaggle competitions where he has 3 gold medals for top positions in leaderboards. As a teammate, he won one Kaggle competition (“Grupo Bimbo Inventory Demand”) where their team proposed the best solution for sales forecasting in the chain of nearly 800 thousand stores.

    Rostyslav Olenchyn

    Business Mentor at FasterCapital, CFO JetSoftPro

    Panel Discussion Language: Ukrainian

    He has a Master's degree in Lviv National University and a Master's degree in Technology Management at Lviv Business School of UCU. His professional interests are business modeling, building strategies for entering the market, fundraising. He has 15 years of experience in building his own business, consulting and anti-crisis management. Rostislav is interested in technology start-ups and their financing. He is a mentor at Creative enterprise UA, the UCU EdTech UCU Enterprise Accelerator and the Fastercapital VC.

    In 2015-2017, he was a member of two projects: DocTravel.com (medical tourism, Seed Forum Kyiv finalist, Seed Forum Oslo 2015), and Kingspeech.ai (EdTech auto self-presentation and remote on-demand training coach) who was a member of the EdTech accelerator. In 2017, he was the mentor in the Hackathon on Big Data, organized by Vodafone Ukraine and the Hackathon of Educational Technologies organized by the Ukrainian Catholic University's Enterprise Center.