Events Calendar

PhD Final Exam – Jose Manuel Picado Leiva

Representationally Robust and Scalable Learning over Relational Databases

Relational databases are the most common means of storing structured data. Learning novel concepts from relational databases is an important problem with applications in several disciplines, such as data management, natural language processing, and bioinformatics. For a learning algorithm to be effective, the input data should be clean and in some desired representation. However, real-world data is usually heterogeneous – the same data may be represented under different representations. For example, one database may contain the value J. Smith, while another database may contain the value John Smith, even though they refer to the same entity. The current approach to effectively use learning algorithms is to find the desired representations for these algorithms, transform the data to these representations, and clean the data. These tasks are hard and time-consuming and are major obstacles for unlocking the value of data. The thesis of this work is that it is possible to develop robust learning algorithms that learn in the presence of representational variations in the data. This approach removes the need for transforming the data before applying learning algorithms. The main contributions of this dissertation are robust learning algorithms that exploit data dependencies to learn directly over databases with representational variations. Data dependencies are usually encoded in the meta-data of the databases or can be extracted from data. We propose several techniques that allow these algorithms to learn efficiently over large databases.

Major Advisor: Arash Termehchy
Committee: Prasad Tadepalli
Committee: Alan Fern
Committee: David Maier
GCR: Hector Vergara Arteaga

Tuesday, May 28, 2019 at 1:30pm to 3:30pm

Weniger Hall, 149
103 SW Memorial Place, Corvallis, OR 97331

Event Type

Lecture or Presentation

Event Topic


Electrical Engineering and Computer Science
Contact Name

Calvin Hughes

Contact Email

Google Calendar iCal Outlook

Recent Activity