Domain Adaptation and Question-Answer Pooling for Aphasia Modelling

MCML Authors

Monica Gonzalez Machorro

→ Group Björn Schuller
 Health Informatics

Björn Schuller

Prof. Dr.

Core PI

Health Informatics

Abstract

In this study, we examine the impact of domain adaptation and question-answer pooling on text-based aphasia prediction with standard and medically specialised BERT models for a German corpus. Modelling tasks comprise<br>aphasia type classification as well as multitask regression of communicative, semantic, and syntactic skills. We found that domain adaptation before finetuning as well as question-answer pooling increased performance for the standard but not for the specialised models on all classification and regression tasks.

inproceedings RGE+25

---

ICNLSP 2025

8th International Conference on Natural Language and Speech Processing. Odense, Denmark, Aug 25-27, 2025.

---

Authors

U. Reichel • M. Gonzalez Machorro • L. Ehlen • P. Hecker • D. Peitz • C. Werner • F. Burkhardt • C. Kohlschein • F. Eyben • B. W. Schuller

---

Links

URL

---

In Collaboration

 Accenture
 Agile Robots
 audEERING

---

Research Area

 B3 | Multimodal Perception