Two publications at Computational Linguistics and Bioinformatics Journals

February 28th, 2025

Great news! Shakila Mahjabin Tonni, our postdoc, has had two new publications accepted at Computational Linguistics, MIT Press and Bioinformatics, Oxford Academic journals. 🥳👏
 
Graded Suspiciousness of Adversarial Texts to Humans
in Computational Linguistics, MIT Press
 
This study explores the challenge of adversarial examples in deep neural networks, focusing on adversarial texts, which differ from adversarial images due to the need for semantic similarity and the discrete nature of text. Unlike image-based adversarial examples, which aim for imperceptibility, adversarial texts must remain non-suspicious to human readers. This research introduce a novel dataset evaluating human perceptions of suspiciousness in adversarial sentences and develops a model to predict suspiciousness levels. This work further examine ways to reduce the detectability of adversarial text by incorporating suspiciousness scores into text generation methods.
 
 
in Bioinformatics, Oxford Academic
 
Enabling clinicians and researchers to directly interact with global genomic data resources by removing technological barriers is vital for medical genomics. AskBeacon enables Large Language Models to be applied to securely shared cohorts via the GA4GH Beacon protocol. By simply “asking” Beacon, actionable insights can be gained and analyzed.