Исследовательская группа

Лаборатория машинного обучения и организации информации

Topics for Student Projects

Byte-addressable Persistent Memory

New storage devices provide much better performance charateristics than rotating drives or flash-based SSDs. However, in order to exploit the full potential of these devices, the database storage techniques must be re-considered.

Holistic Application/Database Tuning

The interaction between applications and databases is often inefficient (too many too tiny queries). this issue cannot be addressed on the DB side only. The goal of this (series of) porjects is to develop techniques for automated application adn database tuning.

Query Processing and Optimization

Declarative query languages are now de-coupled from DBMSs. Well-known query processing and optimization techniques require re-consideration in a broad area of distibuted data processing. Recently introduced approaches, such as to-down optimization and adaptive execution also require further exploration.

Application of Machine Learning to DB Administration Tasks

Several computationally hard problems, such as selection of indexes, selection of materialized views, data partitioning may benefit from application of ML techniques.

Multi-Version Data Structures

The availability of huge volumes of storate enables use of temporal data. To make the processing of temporal data efficient, new data storage techniques should be designed and analyzed.

Machine Learing over Structured Data

Recent publications demonstrate that combining ML with DB techniques may be beneficial in terms of both computational performance and quality of results.

Graph Databases

Several application areas, including semantic Web, Linked data, social networks etc. use graph models, several prototypes of graph databases exist, but hteir performance is often below the expectations.

Students who are interested in any of those subjects are welcomed to contact the Head of Laboratory Boris Asenovich Novikov.