ALIME: Autoencoder Based Approach for Local Interpretability

Abstract

Machine learning and especially deep learning have garnered tremendous popularity in recent years due to their increased performance over other methods. The availability of large amount of data has aided in the progress of deep learning. Nevertheless, deep learning models are opaque and often seen as black boxes. Thus, there is an inherent need to make the models interpretable, especially so in the medical domain. In this work, we propose a locally interpretable method, which is inspired by one of the recent tools that has gained a lot of interest, called local interpretable model-agnostic explanations (LIME). LIME generates single instance level explanation by artificially generating a dataset around the instance (by randomly sampling and using perturbations) and then training a local linear interpretable model. One of the major issues in LIME is the instability in the generated explanation, which is caused due to the randomly generated dataset. Another issue in these kind of local interpretable models is the local fidelity. We propose novel modifications to LIME by employing an autoencoder, which serves as a better weighting function for the local model. We perform extensive comparisons with different datasets and show that our proposed method results in both improved stability, as well as local fidelity.

Machine Learning in Cardiology—Ensuring Clinical Impact Lives Up to the Hype

“…Over time, this may cause a lack of trust in both clinicians unfamiliar with ML principles and patients who are attempting to decide between various therapeutic modalities. 113 Recent guidelines place a substantial focus on enhancing clarity for practitioners, 114 and there are a number of efforts currently underway to help make ML models generally more interpretable, 115 including those utilized within the clinical space. In fact, a whole discipline has been developed that is dedicated to the issue of the interpretability of models for implementation.…”

Explanation of Siamese Neural Networks for Weakly Supervised Learning

“…Numerical results illustrating the explanation method on the MNIST dataset are shown in Figures 9,10,11,12. The figures have the same structure as Figures 5, 6, 7, 8, i.e., every figure contains numerical results of 6 experiments depending on the value of important features s for perturbation.…”

“…The main intuition of LIME is that the explanation may be derived locally from a set of synthetic instances generated randomly in the neighborhood of the instance to be explained such that every synthetic instance has a weight according to its proximity to the explained instance. Several modifications of LIME have been proposed due to success and simplicity of the method, for example, ALIME [12], NormLIME [13], DLIME [14], Anchor LIME [15], LIME-SUP [16], LIME-Aleph [17], SurvLIME [18]. Garreau and Luxburg [19] proposed a thorough theoretical analysis of LIME.…”

Uncertainty Interpretation of the Machine Learning Survival Model Predictions

“…One of the most known methods is LIME [16], which is based on random generation of synthetic examples at a local area around a test example and minimizing the difference between predictions corresponding to these examples provided by an explainable black-box model and an approximating linear model. Following this method, a lot of its modifications were proposed due to simplicity of LIME, for example, ALIME [24], NormLIME [25], DLIME [26], Anchor LIME [27], LIME-SUP [28], LIME-Aleph [29], GraphLIME [30]. The idea of using the linear approximation is also implemented in another method called SHAP [31] which is based on a game-theoretic approach and Shapley values [32].…”