Usman Shahid's profile picture

Usman Shahid

Download resume buttongoogle scholar iconlinkedIn iconGitHub iconEmail icon


I am a Ph.D. student at UIC studying multimodal machine learning. Specifically, I am working towards developing methods that improve knowledge transfer between modalities (co-learning, cross-modal transfer, etc.)

Previously, I have worked on failure point detection in large-scale dialogue systems, frame analysis, causal inference, plagiarism & content spam detection, affective image analysis, sentiment analysis, media analytics, event extraction, and ICTD.

Work Experience

2019 - Now
UIC logo

Research Assistant

Doing my thesis research with Dr. Natalie Parde on multi-modal learning, specifically cross-modal transfer learning and co-learning. Previously, I have worked on medical notes parsing and media bias.

2021 - 2021
Amazon logo

Applied Science Intern

Interned with the Estimated Defect Modeling (EDM) team at Alexa AI. My work entailed building and evaluating neural multi-modal models to automatically identify which component of Alexa failed during a customer interaction.

2018 - 2019
UIC logo

Research Assistant

Conducted research with Dr. Elena Zheleva on model dependence in causal inference from observational network data. In addition to that, I worked on automated moral frame detection in news media.

2016 - 2017
TPI logo

Research Assistant

Besides doing research work on intrinsic plagiarism & content spam detection, media analytics and event mining, I helped create a generic system for visualizing tabular data called Statistan and created an NLP pipeline for automatically annotating news articles called Moodistan.

2015 - 2015
ITU logo

Research Associate

Conducted research on Information and Communication Technologies for Development (ICT4D) and helped develop tools (VillageApps & SpeakMyText) to facilitate underprivileged communities.

Selected Publications

FPI: Failure Point Isolation in Large-scale Conversational Assistants, NAACL'22Rinat Khaziev, Usman Shahid, Tobias Röding, Rakesh Chada, Emir Kapanci, Pradeep Natarajan[PDF]
Detecting and understanding moral biases in news, ACL-NUSE'20Usman Shahid, Barbara Di Eugenio, Andrew Rojecki, Elena Zheleva[PDF] [Data]
Empirical study of model dependence in counterfactual learning from networks, AAAI-WHY'19Usman Shahid, Elena Zheleva[PDF]
High-level concepts for affective understanding of images, WACV'17Afsheen Rafaqat Ali, Usman Shahid, Mohsen Ali, Jeffrey Ho[PDF]
Accurate Detection of Automatically Spun Content via Stylometric Analysis, ICDM'17Usman Shahid, Shehroze Farooqi, Raza Ahmad, Zubair Shafiq, Padmini Srinivasan, Fareed Zaffar[PDF]