Tackling the Accuracy-Interpretability Trade-off: Interpretable Deep Learning Models for Satellite Image-based Real Estate Appraisal
Преодоление компромисса между точностью и интерпретируемостью: интерпретируемые модели глубокого обучения для оценки недвижимости по спутниковым изображениям
2022-10-10
SCID: 54.1/xsdv9c26
Discuss with AI
Grad-Raminterpretable deep learningmulti-view deep learningreal estate appraisalsatellite imagery
Figures from the paper
Abstract (AI)
Deep learning models fuel many modern decision support systems, because they typically provide high predictive performance. Among other domains, deep learning is used in real-estate appraisal, where it allows extending the analysis from hard facts only (e.g., size, age) to also consider more implicit information about the location or appearance of houses in the form of image data. However, one downside of deep learning models is their intransparent mechanic of decision making, which leads to a trade-off between accuracy and interpretability. This limits their applicability for tasks where a justification of the decision is necessary. Therefore, in this article, we first combine different perspectives on interpretability into a multi-dimensional framework for a socio-technical perspective on explainable artificial intelligence. Second, we measure the performance gains of using multi-view deep learning, which leverages additional image data (satellite images) for real estate appraisal. Third, we propose and test a novel post hoc explainability method called Grad-Ram. This modified version of Grad-Cam mitigates the intransparency of convolutional neural networks for predicting continuous outcome variables. With this, we try to reduce the accuracy-interpretability trade-off of multi-view deep learning models. Our proposed network architecture outperforms traditional hedonic regression models by 34% in terms of MAE. Furthermore, we find that the used satellite images are the second most important predictor after square feet in our model and that the network learns interpretable patterns about the neighborhood structure and density.
Key Findings
1
Multi-view deep learning improves real-estate appraisal by incorporating satellite imagery alongside conventional property information.
2
Satellite images are the model’s second most important predictor after square footage, and the network learns interpretable neighborhood structure and density patterns.
3
The paper develops a multidimensional interpretability framework incorporating socio-technical perspectives on explainable artificial intelligence.
4
The proposed Grad-Ram method adapts Grad-Cam to explain convolutional neural networks predicting continuous real-estate values.
5
The proposed architecture outperforms traditional hedonic regression models by 34% in mean absolute error.
Research Object
multi-view deep learning models for satellite image-based real estate appraisal
Research Subject
the trade-off between predictive accuracy and interpretability, including the contribution of satellite imagery and the learned interpretable neighborhood patterns
Publication Details
Publication Date
2022-10-10
Journal
Publisher
ISSN
Access Type
Author Information
Download PDF
Subscribe to digest