About Deep Query
The focus of this project is to utilize recent developments in deep neural networks and associated
models to improve the accuracy and efficientcy of fundamental data and query processing operations.
We are utilizing the ability of certain deep networks to model density in high dimensions. As
such fundamental problems in multi-attribute selectivity estimation and approximate query processing
can be effectively expressed. We have demonstrated orders of magnitude improvement in classic query
processing problems.
These include:
These include:
- Multiattribute selectivity estimations in numerical and categorical data
- Approximate query processing frameworks utilizing deep generative models
- Advanced data acquisition strategies to improve the accuracy of deep models
- Query processing techniques to speed up queries concerning explanations of deep models
- Frameworks for robust deep entity matching and deep entity explanations
Publications
- Zhaoyue Cheng, Nick Koudas, Zhe Zhang and Xiaohui Yu:
Efficient Construction of Non Linear Models Over Normalized Data ICDE 2021 - Sona Hasani, Saravanan Thirumuruganathan, Nick Koudas, Gautam Das:
Shahin: Faster Algorithms for Generating Explanations for Multiple Predictions, SIGMOD 2021 - Yifan Li, Xiaohui Yu, Nick Koudas:
Data Acquisition Strategies for Improving Machine Learning Models, PVLDB 2021 - Suraj Shetiya, Saravanan Thirumuruganathan, Nick Koudas, Gautam Das:
Astrid: Accurate Selectivity Estimation for String Predicates using Deep Learning. Proc. VLDB Endow. 14(4): 471-484 (2020) - Saravanan Thirumuruganathan, Shohedul Hasan, Nick Koudas, Gautam Das:
Approximate Query Processing for Data Exploration using Deep Generative Models. ICDE 2020: 1309-1320 - Shohedul Hasan, Saravanan Thirumuruganathan, Jees Augustine, Nick Koudas, Gautam Das:
Deep Learning Models for Selectivity Estimation of Multi-Attribute Queries. SIGMOD Conference 2020: 1035-1050 - Vincenzo Di Cicco, Donatella Firmani, Nick Koudas, Paolo Merialdo, Divesh Srivastava:
Interpreting deep learning models for entity resolution: an experience report using LIME. aiDM@SIGMOD 2019: 8:1-8:4