Skip to content
conftrace_
2021 EMNLP EMNLP 2021

Interacting Knowledge Sources, Inspection and Analysis: Case-studies on Biomedical text processing

Abstract

AbstractIn this paper we investigate the recently proposed multi-input RIM for inspectability. This framework follows an encapsulation paradigm, where external knowledge sources are encoded as largely independent modules, enabling transparency for model inspection.

🌉 Interdisciplinary Bridge - Artificial Intelligence and Healthcare & Medicine and Natural Language Processing

Authors

Parsa Bagherzadeh, Sabine Bergler

Topics

Artificial Intelligence > Core AI > Interpretability Natural Language Processing > Applications > Information Extraction Healthcare & Medicine > Clinical > Clinical NLP Healthcare & Medicine > Research > Bioinformatics Machine Learning > Bayesian & Probabilistic > Probabilistic Modeling Healthcare & Medicine > Clinical > Medical NLP Natural Language Processing > Applications > Text Processing

Keywords

external knowledge knowledge source text processing biomedical text processing model inspection encapsulation paradigm multi-input rim
Download PDF

More from EMNLP 2021

All EMNLP 2021 papers →
How Familiar Does That Sound? Cross-Lingual Representational Similarity Analysis of Acoustic Word Embeddings
Perturbing Inputs for Fragile Interpretations in Deep Natural Language Processing
An Investigation of Language Model Interpretability via Sentence Editing
Attacks against Ranking Algorithms with Text Embeddings: A Case Study on Recruitment Algorithms
Controlled tasks for model analysis: Retrieving discrete information from sequences
The Acceptability Delta Criterion: Testing Knowledge of Language using the Gradience of Sentence Acceptability