A Robust Unsupervised Ensemble of Feature-Based Explanations using Restricted Boltzmann Machines
arXiv:2111.07379
Abstract
Understanding the results of deep neural networks is an essential step towards wider acceptance of deep learning algorithms. Many approaches address the issue of interpreting artificial neural networks, but often provide divergent explanations. Moreover, different hyperparameters of an explanatory method can lead to conflicting interpretations. In this paper, we propose a technique for aggregating the feature attributions of different explanatory algorithms using Restricted Boltzmann Machines (RBMs) to achieve a more reliable and robust interpretation of deep neural networks. Several challenging experiments on real-world datasets show that the proposed RBM method outperforms popular feature attribution methods and basic ensemble techniques.
References in corpus (16)
- Scikit-learn: Machine Learning in Python
- PyTorch: An Imperative Style, High-Performance Deep Learning Library
- Axiomatic Attribution for Deep Networks
- Learning Important Features Through Propagating Activation Differences
- Deep Neural Networks and Tabular Data: A Survey
- SmoothGrad: removing noise by adding noise
- Captum: A unified and generic model interpretability library for PyTorch
- Sanity Checks for Saliency Maps
- GLocalX -- From Local to Global Explanations of Black Box AI Models
- A Survey on Neural Network Interpretability
- A Deep Learning Approach to Unsupervised Ensemble Learning
- BayLIME: Bayesian Local Interpretable Model-Agnostic Explanations
- On Baselines for Local Feature Attributions
- IROF: a low resource evaluation metric for explanation methods
- Restricted Boltzmann Machines for Robust and Fast Latent Truth Discovery
- Aggregating explanation methods for stable and robust explainability