Publications - Interpretability

SignAttention: On the Interpretability of Transformer Models for Sign Language Translation
Conference paper

This paper studies a Transformer-based model for translating Greek Sign Language to glosses and text. It finds that the model focuses on clusters of frames, with alignment weakening as glosses increase, and shifts from video frames to predicted tokens during translation. This work enhances understanding of SLT models and promotes more transparent translation systems.

Sign Language Translation Transformers Attention Mechanism Greek Sign Language Interpretability

VISTA: Vision Improvement via Split and Reconstruct Deep Neural Network for Fundus Image Quality Assessment
Journal paper

This paper introduces a deep learning model for assessing fundus image quality, essential for diagnosing eye conditions like cataracts and diabetic retinopathy. The model preserves high resolution and includes an autoencoder for reconstruction and classification. Results on the EyeQ dataset show 90.66% accuracy, 88.43% precision, 89.05% recall, and an F1-score of 88.68%.

Autoencoder Network Explainability Fundus Image Gradability Interpretability Quality Assessment Retinal Image

CB-RISE: Improving the RISE Interpretability Method Through Convergence Detection and Blurred Perturbations
Conference paper

This paper enhances the RISE algorithm for image interpretation by introducing C-RISE, which reduces computation time, and CB-RISE, which uses blurred masks instead of black patches. These improvements lead to better heatmap quality and a speedup of about three times.

Black Box Blurred Masks Computer Vision Convergence Detection Deep Learning Interpretability RISE

The Implementation of the RISE Algorithm for the Captum Framework
Conference paper

This paper introduces an efficient implementation of the RISE method in PyTorch using the Captum library. RISE generates explanations for deep learning model predictions by randomly masking input parts. Experiments highlight the significance of the number and size of masks for achieving accurate and versatile interpretations.

Black Box Computer Vision Deep Learning Interpretability RISE

SignAttention: On the Interpretability of Transformer Models for Sign Language TranslationConference paper

VISTA: Vision Improvement via Split and Reconstruct Deep Neural Network for Fundus Image Quality AssessmentJournal paper

CB-RISE: Improving the RISE Interpretability Method Through Convergence Detection and Blurred PerturbationsConference paper

The Implementation of the RISE Algorithm for the Captum FrameworkConference paper

SignAttention: On the Interpretability of Transformer Models for Sign Language Translation
Conference paper

VISTA: Vision Improvement via Split and Reconstruct Deep Neural Network for Fundus Image Quality Assessment
Journal paper

CB-RISE: Improving the RISE Interpretability Method Through Convergence Detection and Blurred Perturbations
Conference paper

The Implementation of the RISE Algorithm for the Captum Framework
Conference paper