NCAF: NTD-based Concept Activation Factorisation Framework for CNN ...
ieeexplore.ieee.org › document
In this study, we rethink the invertible concept-based explainer (ICE) by proposing a non-negative Tucker Decomposition (NTD)based concept activation ...
A non-negative Tucker Decomposition-based concept activation factorization (NCAF) framework for fine-grained categorization explainability is proposed and a ...
In this study, we rethink the invertible concept-based explainer. (ICE) by proposing a non-negative Tucker Decomposition (NTD)- based concept activation ...
Request PDF | On Nov 29, 2023, Ugochukwu Ejike Akpudo and others published NCAF: NTD-based Concept Activation Factorisation Framework for CNN Explainability ...
NTD of tensor A D l in factor matrices W, H,. NCAF: NTD-based Concept Activation Factorisation Framework for CNN Explainability. Conference Paper. Full-text ...
This work studies concept-based explanations and put forward a new definition of concepts as low-dimensional subspaces of hidden feature layers.
Concept activation factorization. We use Non-negative matrix factorization to identify a basis for concepts based on a network's activations (Fig.4).
Missing: NTD- | Show results with:NTD-
Xiaohan Yu, Jun Zhou, Yongsheng Gao NCAF: NTD-based Concept Activation Factorisation Framework for CNN Explainability.IVCNZ 2023: 1-6; 2022. paper. [j4]. Q2.
In this work, we explore three key properties that influence concept activation vectors (CAVs): consistency, entanglement and spatial dependence. First, we ...
Missing: NCAF: NTD-
A detailed survey of explainable deep learning for efficient and robust pattern recognition is represented.