Розклад конференції: Лекції

Download lecture materials

Mathematics is old, Logic is old, but in some sense, many basic ideas of the foundation of computer science are old too. For example, that proof checking (or type checking) is decidable but proof construction (or type inference) is not, was hinted at by Aristotle. In this talk, I go through some of the developments in mathematics and logic which influenced the creation of some computer science ideas in the 20th century. In particular, I discuss how the need for more precision and formality in the 18th century, led to the development of logic in the 19th century, to the work of Frege and the discovery of Russell's paradox. I then discuss the use of type theory (a concept that was already implicit in Euclid's Geometry 325 B.C.) by Russell to avoid the paradox and explain the development and the influence of types in computation.
Download lecture materials

In this lecture, I will focus on interrelations between security, privacy and trust and consider some technological approaches suitable for privacy protection. I will consider sources of privacy threats and motivation for protection. Particularly, I will focus on potential ability of secure multi-party computations to provide privacy protection. I will demonstrate it via examples of such solutions starting from some classical examples and move to more efficient solutions published recently.
Download lecture materials

Download lecture materials

Report devoted to the changes that are scheduled in Java 8: Support lambda-function, expansion of interface default-methods, evolution of interface Collection, support parallel operations with collections at libraries.
Download lecture materials

In this talk I will explain how structural properties of natural languages can be used to decrease the complexity of a document classification task in order to make it more tractable for an unsupervised machine learning algorithm. I will focus on three fundamental design features that deal mostly with word order phenomena and can therefore be easily formalized. I will further show how the strategies based on these features also allow to handle noisy data and unseen examples efficiently, as is often necessary when dealing with natural language data. The presentation will include a relevant case study to address the main points. The following topics will be covered:

• Basic concepts of unsupervised machine learning for document classification
• Characterization of linguistic inputs from social media data
• Using linguistically motivated units of analysis for computational tractability
• Using a non-configurational grammatical formalism in order to capture syntactic relations
• Leveraging (linguistic) transitivity to infer semantic relations
• Customized knowledge-acquisition through unsupervised classification

Download lecture materials

In the talk I will describe the three key pieces of practical NLP work: problem formulation, working with data, and NLP tools. I will also highlight those aspects of them, that are influenced by engineering and business requirements, and are not directly related to academic research, but nevertheless are important to bringing theoretical results into life.
The following topics will be discussed:

• Business requirements in NLP problems and how to deal with them - Differences between theoretical and practical NLP
• Getting data
• Approaches to creating and/or improving data
• Collecting data from the users
• Handling huge amounts of data
• Selecting appropriate tools
• 3 levels of linguistic tools
• Working with 3rd-party APIs

A number of theoretical approaches to solve the problem of 3D-packing exist, but practical solutions have not been studied widely mostly because few industries need this. 3D printing is one of such young industries that motivates new solutions to be discovered, implemented, compared and integrated into real world, and Materialise has been the pioneer and remains the leader of related markets. We will share our experience of applying different space optimization algorithms to problem of reducing costs of printing huge volumes of 3D parts applied to Medical and Automotive directions of our businesses.
Download lecture materials