Quality In, Quality Out: Ensuring the Quality of Data Inputs for Intelligent Systems
Abstract
Modern software systems (e.g., AI-enabled systems) depend on data to provide intelligent services to customers. Garbage in, garbage out: the quality of input data significantly affects the quality of the provided services. In this talk, I will share recent work in the MOOSE lab that aims to provide infrastructural support to improve the data quality for downstream tasks. This talk will discuss solutions to handle unstructured data, structured data, and data shifts. The first part of the talk will discuss a solution that converts unstructured data (e.g., software logs) into structured data; the second part of the talk will discuss a solution that automatically derives patterns from structured data and detects data anomalies; finally, the talk will discuss the challenges of dealing with data shifts.
Heng Li
Assistant Professor - Polytechnique Montreal
Heng Li is an assistant professor in the Department of Computer and Software Engineering at Polytechnique Montreal, where he leads the MOOSE lab. He holds a PhD in Computing from Queen’s University. Prior to his academic career, he worked in the industry for years as a software engineer at Synopsys and as a software performance engineer at BlackBerry. His and his students’ research aims to address the practical challenges in software monitoring, software quality engineering, intelligent operations of software systems, and quality engineering of machine learning applications. The outcomes of such research have benefited the daily maintenance and operations of software and software-intensive systems in the industry and inspired follow-up research in related areas. He is a core organizer of the international Software Engineering for Machine Learning Applications (SEMLA) symposium. He is a recipient of the Discovery Grant from NSERC, John R. Evans Leaders Fund from CFI, and NSERC Alliance, among other grants. He is the secretary of the Standard Performance Evaluation Corporation (SPEC) Research Group – DevOps Performance Working Group.