Layer factor analysis in convolutional neural networks for explainability
dc.contributor.author | López González, Clara Isabel | |
dc.contributor.author | Gómez Silva, María José | |
dc.contributor.author | Besada Portas, Eva | |
dc.contributor.author | Pajares Martínsanz, Gonzalo | |
dc.date.accessioned | 2024-02-02T15:38:06Z | |
dc.date.available | 2024-02-02T15:38:06Z | |
dc.date.issued | 2024-01 | |
dc.description | Se trata de un acuerdo transformativo. | |
dc.description.abstract | Explanatory methods that focus on the analysis of the features encoded by Convolutional Neural Networks (CNNs) are of great interest, since they help to understand the underlying process hidden behind the black-box nature of these models. However, to explain the knowledge gathered in a given layer, they must decide which of the numerous filters to study, further assuming that each of them corresponds to a single feature. This, coupled with the redundancy of information, makes it difficult to ensure that the relevant characteristics are being analyzed. The above represents an important challenge and defines the aim and scope of our proposal. In this paper we present a novel method, named Explainable Layer Factor Analysis for CNNs (ELFA-CNNs), which models and describes with quality convolutional layers relying on factor analysis. Regarding contributions, ELFA obtains the essential underlying features, together with their correlation with the original filters, providing an accurate and well-founded summary. Through the factorial parameters we gain insights about the information learned, the connections between channels, and the redundancy of the layer, among others. To provide visual explanations in a similarly way to other methods, two additional proposals are made: a) Essential Feature Attribution Maps (EFAM) and b) intrinsic features inversion. The results prove the effectiveness of the developed general methods. They are evaluated in different CNNs (VGG-16, ResNet-50, and DeepLabv3+) on generic datasets (CIFAR-10, imagenette, and CamVid). We demonstrate that convolutional layers adequately fit a factorial model thanks to the new metrics presented for factor and fitting residuals (D1, D>, and Res, derive from covariance matrices). Moreover, knowledge about the deep image representations and the learning process is acquired, as well as reliable heat maps highlighting regions where essential features are located. This study effectively provides an explainable approach that can be applied to different CNNs and over different datasets. | |
dc.description.department | Depto. de Ingeniería de Software e Inteligencia Artificial (ISIA) | |
dc.description.department | Sección Deptal. de Arquitectura de Computadores y Automática (Físicas) | |
dc.description.faculty | Fac. de Informática | |
dc.description.faculty | Fac. de Ciencias Físicas | |
dc.description.refereed | TRUE | |
dc.description.sponsorship | Comunidad Autónoma de Madrid | |
dc.description.sponsorship | Ministerio de Ciencia e Innovación (España) | |
dc.description.sponsorship | Unión Europea NextGeneration | |
dc.description.sponsorship | Ministerio de Universidades (España) | |
dc.description.status | pub | |
dc.identifier.citation | López-González CI, Gómez-Silva MJ, Besada-Portas E, Pajares G. Layer factor analysis in convolutional neural networks for explainability. Applied Soft Computing. 2024 Jan;150:111094-111 | |
dc.identifier.doi | 10.1016/j.asoc.2023.111094 | |
dc.identifier.issn | 1568-4946 | |
dc.identifier.officialurl | https://www.sciencedirect.com/science/article/pii/S1568494623011122 | |
dc.identifier.uri | https://hdl.handle.net/20.500.14352/98403 | |
dc.journal.title | Applied Soft Computing | |
dc.language.iso | eng | |
dc.page.final | 111111 | |
dc.page.initial | 111094 | |
dc.publisher | Elsevier | |
dc.relation.projectID | info:eu-repo/grantAgreement/CAM/PRICIT/Y2020/TCS-6420//Hacia un sistema Integral para la Alerta y Gestión de BLOOMs de cianobacterias en aguas continentales/IA-GES-BLOOM-CM | |
dc.relation.projectID | info:eu-repo/grantAgreement/MCIN/AEI/10.13039/501100011033/EU/PRTR/TED2021-130123B-I00//Más allá del uso de tecnologías digitales en blooms de cianobacterias: gestión inteligente de cianobacterias mediante el uso de gemelos digitales y computación en el borde/SMART-BLOOMS | |
dc.relation.projectID | info:eu-repo/grantAgreement/MCIN/PEICTI/PID2021-127648OB-C33/ES/Cooperación de vehículos de superficie y aéreos para aplicaciones de inspección en entornos cambiantes/INSERTION | |
dc.rights | Attribution 4.0 International | en |
dc.rights.accessRights | open access | |
dc.rights.uri | http://creativecommons.org/licenses/by/4.0/ | |
dc.subject.cdu | 004.8 | |
dc.subject.cdu | 004.85 | |
dc.subject.cdu | 004.932 | |
dc.subject.cdu | 004.032.26 | |
dc.subject.keyword | Deep learning | |
dc.subject.keyword | Explainable Artificial Intelligence (xAI) | |
dc.subject.keyword | Statistical modeling | |
dc.subject.keyword | Visual explanation | |
dc.subject.keyword | Feature learning | |
dc.subject.keyword | Attribution map | |
dc.subject.ucm | Inteligencia artificial (Informática) | |
dc.subject.unesco | 1203.17 Informática | |
dc.title | Layer factor analysis in convolutional neural networks for explainability | |
dc.type | journal article | |
dc.type.hasVersion | VoR | |
dc.volume.number | 150 | |
dspace.entity.type | Publication | |
relation.isAuthorOfPublication | 779a7137-78a8-46a7-81e0-58b8bd5f1748 | |
relation.isAuthorOfPublication | 0acc96fe-6132-45c5-ad71-299c9dcb6682 | |
relation.isAuthorOfPublication | 878e090e-a59f-4f17-b5a2-7746bed14484 | |
relation.isAuthorOfPublication.latestForDiscovery | 779a7137-78a8-46a7-81e0-58b8bd5f1748 |
Download
Original bundle
1 - 1 of 1
Loading...
- Name:
- Layer factor analysis in cnn for explainability.pdf
- Size:
- 8.92 MB
- Format:
- Adobe Portable Document Format