The Computational Language Understanding (CLU) Lab is principally concerned with advancing computational models to understand the semantics of human language and translating these models into clinical practice. Our approach is to seek a good understanding of human language, to capture the essential features of language in appropriate models, and to develop the necessary computational framework to advance knowledge and discovery in Natural Language Processing (NLP) and Healthcare. We accomplish this through rigorous computational research coupled with partnerships with linguists and clinicians.

Members

Hadi Amiri
Assistant Professor

Mohamed Elgaar
PhD student, F'20

Nidhi Vakil
PhD student, S'21

Jiali Cheng
PhD student, S'23

Research

Several specific areas that we are currently investigating are listed below:

• Curriculum Learning for NLP: Deep neural networks can effectively tackle many NLP tasks, but they could be computationally expensive to train. How can we uncover the salient characteristics of these learners (networks) and their learning materials (training data) for effective representation and efficient training?

• Clinical Decision Support: Medical information in referral letters, physician notes or scientific articles are locked in unstructured text. What are the best techniques to extract insight from such data, represent patient data and reference materials about diseases, triage patient applications, pinpoint disease-causing gene variants, and enhance clinical decision support systems with evidence?

• Social Media Surveillance: User generated content in social media present naturally occurring data that can be used to obtain low-cost and high-resolution views into population behavior. How can we develop online surveillance systems that can monitor population behavior at scale to detect (health-related) trends and outbreaks, and identify opportunities for decision making or intervention?

See further details here.

Recent Publications

• Controlled Transformation of Text-Attributed Graphs
  Nidhi Vakil, Hadi Amiri. In Proceedings of The 2024 Conference on Empirical Methods in Natural Language Processing (EMNLP’24) (Findings).

• FairFlow: Mitigating Dataset Biases through Undecided Learning for Natural Language Understanding
  Jiali Cheng, Hadi Amiri. In Proceedings of The 2024 Conference on Empirical Methods in Natural Language Processing (EMNLP’24).

• MedDec: A Dataset for Extracting Medical Decisions from Discharge Summaries
  Mohamed Elgaar, Jiali Cheng, Nidhi Vakil, Hadi Amiri, Leo Anthony Celi. In Findings of the Association for Computational Linguistics (ACL’24) (Findings). [PDF]

• CogniVoice: Multimodal and Multilingual Fusion Networks for Mild Cognitive Impairment Assessment from Spontaneous Speech
  Jiali Cheng, Mohamed Elgaar, Nidhi Vakil, Hadi Amiri. In Proceedings of INTERSPEECH 2024 (INTERSPEECH’24). [PDF]

• MultiDelete for Multimodal Machine Unlearning
  Jiali Cheng, Hadi Amiri. In Proceedings of The 18th European Conference on Computer Vision (ECCV’24). [PDF]

• Ling-CL: Understanding NLP Models through Linguistic Curricula
  Mohamed Elgaar, Hadi Amiri. In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing (EMNLP’23). [PDF]

• Complexity-Guided Curriculum Learning for Text Graphs
  Nidhi Vakil, Hadi Amiri. In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing (EMNLP’23) (Findings). [PDF]

• HuCurl: Human-induced Curriculum Discovery
  Mohamed Elgaar, Hadi Amiri. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (ACL’23). [PDF]

• Curriculum Learning for Graph Neural Networks: A Multiview Competence-based Approach
  Nidhi Vakil, Hadi Amiri. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (ACL’23). [PDF]

See the complete list here.

Contact

• Address: One University Avenue, 320 Southwick Hall, Lowell, MA 01854

• Email: clu@cs.uml.edu