Explaining Deep Learning Models for Tabular Data Using Layer-Wise Relevance Propagation
Trust and credibility in machine learning models are bolstered by the ability of a model to explain its decisions. While explainability of deep learning models is a well-known challenge, a further challenge is clarity of the explanation itself for relevant stakeholders of the model. Layer-wise Relev...
Main Authors: | , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
MDPI AG
2021-12-01
|
Series: | Applied Sciences |
Subjects: | |
Online Access: | https://www.mdpi.com/2076-3417/12/1/136 |
_version_ | 1797499662384496640 |
---|---|
author | Ihsan Ullah Andre Rios Vaibhav Gala Susan Mckeever |
author_facet | Ihsan Ullah Andre Rios Vaibhav Gala Susan Mckeever |
author_sort | Ihsan Ullah |
collection | DOAJ |
description | Trust and credibility in machine learning models are bolstered by the ability of a model to explain its decisions. While explainability of deep learning models is a well-known challenge, a further challenge is clarity of the explanation itself for relevant stakeholders of the model. Layer-wise Relevance Propagation (LRP), an established explainability technique developed for deep models in computer vision, provides intuitive human-readable heat maps of input images. We present the novel application of LRP with tabular datasets containing mixed data (categorical and numerical) using a deep neural network (1D-CNN), for Credit Card Fraud detection and Telecom Customer Churn prediction use cases. We show how LRP is more effective than traditional explainability concepts of Local Interpretable Model-agnostic Explanations (LIME) and Shapley Additive Explanations (SHAP) for explainability. This effectiveness is both local to a sample level and holistic over the whole testing set. We also discuss the significant computational time advantage of LRP (1–2 s) over LIME (22 s) and SHAP (108 s) on the same laptop, and thus its potential for real time application scenarios. In addition, our validation of LRP has highlighted features for enhancing model performance, thus opening up a new area of research of using XAI as an approach for feature subset selection. |
first_indexed | 2024-03-10T03:50:38Z |
format | Article |
id | doaj.art-c86d6d88f5c94af4ae9f656d1cbfd10a |
institution | Directory Open Access Journal |
issn | 2076-3417 |
language | English |
last_indexed | 2024-03-10T03:50:38Z |
publishDate | 2021-12-01 |
publisher | MDPI AG |
record_format | Article |
series | Applied Sciences |
spelling | doaj.art-c86d6d88f5c94af4ae9f656d1cbfd10a2023-11-23T11:08:16ZengMDPI AGApplied Sciences2076-34172021-12-0112113610.3390/app12010136Explaining Deep Learning Models for Tabular Data Using Layer-Wise Relevance PropagationIhsan Ullah0Andre Rios1Vaibhav Gala2Susan Mckeever3CeADAR Irelands Center for Applied AI, University College Dublin, D04V2N9 Dublin, IrelandCeADAR Irelands Center for Applied AI, Technological University Dublin, D07ADY7 Dublin, IrelandCeADAR Irelands Center for Applied AI, Technological University Dublin, D07ADY7 Dublin, IrelandCeADAR Irelands Center for Applied AI, Technological University Dublin, D07ADY7 Dublin, IrelandTrust and credibility in machine learning models are bolstered by the ability of a model to explain its decisions. While explainability of deep learning models is a well-known challenge, a further challenge is clarity of the explanation itself for relevant stakeholders of the model. Layer-wise Relevance Propagation (LRP), an established explainability technique developed for deep models in computer vision, provides intuitive human-readable heat maps of input images. We present the novel application of LRP with tabular datasets containing mixed data (categorical and numerical) using a deep neural network (1D-CNN), for Credit Card Fraud detection and Telecom Customer Churn prediction use cases. We show how LRP is more effective than traditional explainability concepts of Local Interpretable Model-agnostic Explanations (LIME) and Shapley Additive Explanations (SHAP) for explainability. This effectiveness is both local to a sample level and holistic over the whole testing set. We also discuss the significant computational time advantage of LRP (1–2 s) over LIME (22 s) and SHAP (108 s) on the same laptop, and thus its potential for real time application scenarios. In addition, our validation of LRP has highlighted features for enhancing model performance, thus opening up a new area of research of using XAI as an approach for feature subset selection.https://www.mdpi.com/2076-3417/12/1/136explainability1D-CNNstructured datalayer-wise relevance propagationdeep learningtransparency |
spellingShingle | Ihsan Ullah Andre Rios Vaibhav Gala Susan Mckeever Explaining Deep Learning Models for Tabular Data Using Layer-Wise Relevance Propagation Applied Sciences explainability 1D-CNN structured data layer-wise relevance propagation deep learning transparency |
title | Explaining Deep Learning Models for Tabular Data Using Layer-Wise Relevance Propagation |
title_full | Explaining Deep Learning Models for Tabular Data Using Layer-Wise Relevance Propagation |
title_fullStr | Explaining Deep Learning Models for Tabular Data Using Layer-Wise Relevance Propagation |
title_full_unstemmed | Explaining Deep Learning Models for Tabular Data Using Layer-Wise Relevance Propagation |
title_short | Explaining Deep Learning Models for Tabular Data Using Layer-Wise Relevance Propagation |
title_sort | explaining deep learning models for tabular data using layer wise relevance propagation |
topic | explainability 1D-CNN structured data layer-wise relevance propagation deep learning transparency |
url | https://www.mdpi.com/2076-3417/12/1/136 |
work_keys_str_mv | AT ihsanullah explainingdeeplearningmodelsfortabulardatausinglayerwiserelevancepropagation AT andrerios explainingdeeplearningmodelsfortabulardatausinglayerwiserelevancepropagation AT vaibhavgala explainingdeeplearningmodelsfortabulardatausinglayerwiserelevancepropagation AT susanmckeever explainingdeeplearningmodelsfortabulardatausinglayerwiserelevancepropagation |