Accelerating Data Exploration
Students: David Solodukhin, Daniel Pagan, Brian Cai, and Apaar ShankerExploratory statistical analysis of data requires repeated aggregation of data that precludes interactive analysis. We propose techniques for caching the results of queries to accelerate exploratory statistical analysis.
Accelerating DDL Operations
Students: Teju Nareddy, Grace Harper, and Sai GundlapalliSchema changes often require large amounts of data to be moved from table to table, thereby straining system resources and increasing the likelihood of errors. We seek to design a faster DDL implementation that allows for near-instantaneous schema modifications while reducing the run-time impact on subsequent queries.
VQL: Video Query Language
Students: Kedi Zheng, Minyi Hu, and Vijay sri LakshmananWe propose a high-level declarative language for video analytics. VQL queries operate on spatio-temporal features of objects contained in videos.
Adapting ML Models to Concept Drift
Students: Abhijit Suprem and Rodrigo Alves LimaWe present a system for concept drift adaptation on evolving streams of data without limiting assumptions on the drift in data. We place our drift adaptation in context of event detection on web-data based social sensors.
Learning Database Management Protocols via JavaScript Demos (I)
Students: Mark Woodson, Leo Weng, and Xi ChengWe design student-friendly JavaScript demos that are geared towards learning database management protocols.
Video Analytics using Action Recognition
Students: Pramod Chunduri, Yanan Wang, Zhanhao Liu, and Yuhong WangWe develop a framework for efficiently querying on actions contained in a video by using context from multiple frames.
Reliable Persistent Data-Structures
Students: Pradeep Fernando and Jiashen CaoWe propose a reliable persistent storage stack based on non-volatile memory and modern networking system software.
Quantifying State Staleness
Students: Harshit Gupta, Salini Mishra, and Lucia Verdejo EstevezWe present a scheme for coordinated management of multiple clusters via state synchronization. We formulate a model for the same that takes into account various factors like overhead for state synchronization, degree of staleness for clusters and overall time taken for the complete process.
Learning Database Management Protocols via JavaScript Demos (II)
Students: Jeremy Peterson and Vivian ThiebautWe design student-friendly JavaScript demos that are geared towards learning database management protocols.
BuzzDB (I)
Students: Huajun Guo, Jonathan Lee, and Binze CaiWe port a multi-user transactional database server written in Java to C++ for pedagogical purposes.
BuzzDB (II)
Students: Alonzo Hernandez, Jenny Li, and Omega HaileyesusWe port a multi-user transactional database server written in Java to C++ for pedagogical purposes.
Video Storage Management
Students: Jaeho Bang, Shreya Dubey, and Ujjwal AroraWe design a storage manager for accelerating video analytics.
Query Equivalence
Students: Qi Zhou and David HeathWe propose an automated technique for detecting the equivalence of SQL queries. By automatically accounting for overlapping computation, it reduces computational resources expended on query processing.