معرفی
Dr. Jelle Zuidema is Associate Professor in Natural Language Processing, Explainable AI and Cognitive Modelling at the Institute for Logic, Language and Computation (ILLC) of the University of Amsterdam, with primary affiliation in the Natural Language Processing research group and secondary affiliation in Language & Music Cognition. He directs the Cognition, Language and Computation lab (CLC-lab) within the Faculty of Science.
His research bridges artificial intelligence, cognitive science and linguistics with focus on interpretable deep learning models for text and sound. He pioneered techniques including diagnostic classification (probing), attention rollout, TreeLSTM, and masked language modeling. His work addresses fundamental questions about hierarchical compositionality in language and how neural networks can represent linguistic structure.
Analysis of his publication record reveals consistent innovation in neural network interpretability methods, with recent work focusing on bias detection in language models and attention flow quantification. His research trajectory shows evolution from foundational neural parsing work to current emphasis on explainable AI and cognitive plausibility of deep learning models.
Dr. Zuidema actively supervises PhD and MSc students while teaching courses including Evolution of Language and Music, Foundations of Neural and Cognitive Modelling, and Interpretability & Explainability in AI. He coordinates the Cognitive Science track in the Brain & Cognitive Sciences master's program.
He directs the CLC-lab which pursues both hypothesis-driven methods (diagnostic classifiers, representational similarity analysis) and data-driven methods (layer-wise relevance propagation, contextual decomposition) for neural network interpretation. The lab's InDeep project focuses on interpreting deep learning models for text and sound processing.
