Duration: 2h

World's data is growing exponentially. To take control over it, companies were started using modern cluster-based solutions, currently, Map-Reduce is a "state of the art" approach for analyzing large amounts of data. This course will give you intuition into common Map-Reduce workflows and tasks. The main tool during this class will be open-source implementation of Map-Reduce - Apache Hadoop and more high-level query language Apache Pig.

• Introduction into Map-Reduce algorithm
• Common tasks for Map-Reduce
• Hadoop infrastructure
• From Hadoop to Pig
• Map-Reduce practice

Materials of the course

Tutor

Mr. Stanislav Levental

Country: Ukraine

Place of employment: Grammarly Inc., Kiev

Affiliation: Data Engineer

Spheres of interests: big data, statistics, NLP, databases

Email: This email address is being protected from spambots. You need JavaScript enabled to view it.

Duration: 3 h

Why we should visualize things? Because our image processing ability at least ten times more powerful then for example processing of texts. Because frequently it's the shortest path from data to knowledge. So in this short course we'll se a couple of examples of best visualizations and we'll learn some underlying principles of information design. Last, but not least, some of the most useful tools and libraries will be introduced to our listeners.

The course consist of:

• general introduction - why is visualization important - best examples
• shortcuts: some canons for effective information design
• introduction to some tools
• where to go further

Materials of the course

Tutor

Mr. Anatoly Bondarenko

Place of employment: data journalism editor, Texty.org.ua

E-mail: This email address is being protected from spambots. You need JavaScript enabled to view it.

During a lecture and a practice session, the students will be presented with the introduction to the basic concepts of data management, such as relational databases (SQL), key-value storage (NoSQL), multidimensional databases (array storage, MDX, OLAP), graph databases, directory services, and metadata formats.

Materials of the course

Tutor

Mr. Konstantin Tretyakov

Country: Estonia

Place of employment: University of Tartu, Institute of Computer Science

Email:This email address is being protected from spambots. You need JavaScript enabled to view it.

Course duration: 10-12 h

Prerequisites: general knowledge of Algorithms and Data Structures, Statistics

Natural language processing consists of encoding knowledge in formal models. However, in addition to encoding explicit language-specific knowledge (e.g. “маленька парасолька”UKR means “small umbrella”ENG) one can also encode meta-knowledge: assumptions on what kind of knowledge there is (e.g. “phrases in Ukrainian have corresponding translations in English”), and then use these assumptions (e.g. “phrases and their translations are likely to co-occur”) to automatically discover language-specific knowledge in recorder examples of language usage (text corpora) without having to specify it yourself. This approach is called unsupervised language learning and it will be the central topic of this course.

Topics:

• unsupervised morphological segmentation
• unsupervised parsing
• word and phrase alignment
• statistical machine translation
• text clustering

Materials of the course

Tutor

Dr. Mark Fishel

Country: Estonia/Switzerland

Place of employment: Post-doc, Institute of Computational Linguistics, University of Zurich

Curriculum Vitae: PhD in Computer Science from the University of Tartu, currently a post-doc in statistical machine translation at the University of Zurich.

Spheres of interests: machine translation, machine learning, computational natural language learning

E-mail: This email address is being protected from spambots. You need JavaScript enabled to view it.

Course duration: 8 h 

Natural Language Processing (NLP) exists at the intersection of 3 disciplines: Computer Science, Linguistics, and Software Engineering. It is based on the foundation of Algorithms, Machine Learning, and Statistics, and is aimed at solving a wide range of real-world computational problems related to human cognition, such as: machine translation, text classification, automated question answering, error correction, etc. We will learn the foundational models of text representation, explore some of the most important algorithms, and work with the tools for applying these approaches to real-world data sets. The course will be built around developing a solution to a particular practical NLP problem. The students will have a choice of 3 languages for performing practical tasks, the code in the course will be presented in Common Lisp (a brief introduction to the language will be given).
Prerequisites: general knowledge of Algorithms and Data Structures, and Statistics

Topics:

• linguistic concepts in NLP
• most important NLP problems and domains
• working with corpora and real-world data sets
• language models and corresponding generative learning algorithms: bag-of-words model (Bayesian learning), sequence models (hidden Markov models), structural models, holistic models (artificial neural networks)
• discriminative learning algorithms: Maximum entropy, Decision Trees, etc.
• practical applications: Bayesian document classification, CKY parsing, Noisy Channel spelling correction, etc.

Required software (free choice of one of the following options):

• Python NLTK (recommended reading: TBA)
• Stanford Core NLP (recommended reading: TBA)
• Common Lisp cl-nlp (recommended reading: TBA)

Materials of the course

Tutor

Mr. Vsevolod Dyomkin

Country: Ukraine

Place of employment: Grammarly Inc.

Spheres of research: error correction

E-mail: This email address is being protected from spambots. You need JavaScript enabled to view it.

Phone: +38 096 111 41 56

Course duration: 2 h

This course will show and explain different approaches and tricks how to efficiently store and process large amounts of different data. Need to blazingly fast processing terabytes of satellite imagery or billions of data points in time series data — this course will help you.

Topics:

• Find and remove bottleneck. Repeat.
• Physical restrictions of hardware and how to deal with this.
• Rethinking how you use data and store data accordingly.
• Strong and weak sides of SQL Databases. What’s going under hood.
• NoSQL into SQL Database.

Materials of the course

Tutor

Mr. Oleksii Leonov

Country: Ukraine

Place of employment: https://cropio.com

Curriculum Vitae: M.Sc., System analysis from the National University of Ukraine "KPI"

Affiliation: Chief Technology Officer at Cropio

E-mail: This email address is being protected from spambots. You need JavaScript enabled to view it.

Course duration: 4 h

This course will introduce you to satellite imagery and give basic experience with satellite images processing.
Pre-requisites: basics of Python needed to fully understand practical part of the course. Basics of linux/unix console recommended.

Topics:

• How satellites take images from space and what kind of data we can obtain.
• How this data could be used in economics, agriculture, ecology and other types of science and business researches.
• Where you can get free raw satellite imagery archives and how to process this raw data.
• Tools for satellite imagery processing (QGIS, OGR and GDAL packages and their Python API).

Materials of the course

Tutor

Mr. Oleksii Leonov

Country: Ukraine

Place of employment: https://cropio.com

Curriculum Vitae: M.Sc., System analysis from the National University of Ukraine "KPI"

Affiliation: Chief Technology Officer at Cropio

E-mail: This email address is being protected from spambots. You need JavaScript enabled to view it.

Duration: 8-10 h

This course is intended for students, who are curious about data science and machine learning, but do not have deep background in mathematics. Main focus of this introductory workshop is various statistical-learning techniques with the predictive purpose - in other words, the fascinating field of “supervised machine learning”. This course will help participants to gain an intuition about general methods in the field and apply them on practice. We will discuss how to choose proper methods and prepare data to avoid fallacies, how to build, improve and compare models, and at last but not the least, how to understand them and “tell the story” of your results. We will be using an open source software R that is one of the most popular analysis tools with the broad “fan-zone”.

Materials of the course

Tutor

Ms. Anna Leontjeva

Country: Estonia

Place of employment: Software Technology and Applications Competence Center, Tartu, Estonia

Affiliation: researcher

Spheres of interests: machine learning, data mining, applied statistics, statistical learning

Email: This email address is being protected from spambots. You need JavaScript enabled to view it.

Duraction: 8 h

The aim of the course is to make students familiar to the fundamentals of the speech recognition technology. We will learn how to extract informative features from speech, how to model speech dynamics with hidden Markov models, and how to build language models corresponding to the task domain. A brief historical overview of the speech recognition methods as well as the future trends will be given. During the practical part of the course students will be proposed to develop a basic speech recognizer using modern open-source software.

Pre-requisites: general knowledge on machine learning, neural networks (optional).

Methods: Machine Learning & Pattern Recognition -> Supervised Learning:

• Hidden Markov models;
• Gaussian mixture models;
• Neural networks;
• Language modeling techniques: grammars (JSGF), N-grams.

Materials of the course

Tutor

Mr. Dmytro Prylipko

Country: Germany

Place of employment: Otto-von-Guericke University Magdeburg, Germany

Spheres of research: affect recognition from speech, modeling non-linguistics vocalizations in spontaneous speech, recognition of spontaneous speech, spoken dialog systems design

Email:This email address is being protected from spambots. You need JavaScript enabled to view it.

Phone: +49 176 325 44 562

Recommender systems are everywhere on the Web these days: they help us find friends in social networks, they recommend us items when we shop online, they manage our personalized internet radio and assist us every time we search something online. In this course we discuss three state-of-the-art approaches to build recommender systems: content based recommenders, memory-based collaborative filtering and random walks on graphs. Memory-based collaborative filtering is the most commonly used approach for e-commerce systems which looks for similarities in users’ purchasing histories. It relies on an assumption that users who have purchased common items in the past would buy similar things in future and recommends new items from peers with similar purchasing histories. In contrast, content-based recommenders focus on content features. Typically, a supervised learning model is trained to recognize content features which users have liked before and is further applied to predict users’ interests in new content items. Random-walks is an alternative approach which operates on linked data represented with graphs (e.g., graph of HTML links, social network graph, etc.) and is commonly used to rank web pages in search engines and recommend friends in social networks.

Keywords: recommender systems, collaborative filtering, content-based recommender systems, random walks with restart

Pre-requisites: basic knowledge in

• linear algebra
• graph theory
• supervised learning (optional)

Tools: GraphLab, SciPy

Materials of the course

Tutor

Dr. Dmytro Karamshuk

Country: Ukraine, United Kingdom

Affiliation: King’s College London, Department of Informatics

Curriculum Vitae: I got an MSc. in computer science and engineering from the National University of Ukraine "KPI" in 2007. From 2005 to 2010 I worked at a co-founded software company Stanfy LLC. I hold a PhD in computer science from IMT Institute for Advanced Studies, Lucca where I worked in collaboration with the Institute for Informatics and Telematics of National Research Council (IIT-CNR), Pisa. In 2012 I was a visiting researcher at the Computer Laboratory, University of Cambridge. Since 2013 I’m a postdoc at King's College London.

Research interests: data mining, complex networks, machine learning, human mobility, mobile networks

Email: This email address is being protected from spambots. You need JavaScript enabled to view it.