Meike Nauta

Third-generation artificial neural networks, Spiking Neural Networks (SNNs), can be efficiently implemented on hardware. Their implementation on neuromorphic chips opens a broad range of applications, such as machine learning-based autonomous control and intelligent biomedical devices. In critical applications, however, insight into the reasoning of SNNs is important, thus SNNs need to be equipped with the ability to explain how decisions are reached. We present Temporal Spike Attribution (TSA), a local explanation method for SNNs. To compute the explanation, we aggregate all information available in model-internal variables: spike times and model weights. We evaluate TSA on artificial and real-world time series data and measure explanation quality w.r.t. multiple quantitative criteria. We find that TSA correctly identifies a small subset of input features relevant to the decision (i.e., is output-complete and compact) and generates similar explanations for similar inputs (i.e., is continuous). Further, our experiments show that incorporating the notion of absent spikes improves explanation quality. Our work can serve as a starting point for explainable SNNs, with future implementations on hardware yielding not only predictions but also explanations in a broad range of application scenarios. Source code is available at https://github.com/ElisaNguyen/tsa-explanations.

Interpretable part-prototype models are computer vision models that are explainable by design. The models learn prototypical parts and recognise these components in an image, thereby combining classification and explanation. Despite the recent attention for intrinsically interpretable models, there is no comprehensive overview on evaluating the explanation quality of interpretable part-prototype models. Based on the Co-12 properties for explanation quality as introduced in arXiv:2201.08164 (e.g., correctness, completeness, compactness), we review existing work that evaluates part-prototype models, reveal research gaps and outline future approaches for evaluation of the explanation quality of part-prototype models. This paper, therefore, contributes to the progression and maturity of this relatively new research field on interpretable part-prototype models. We additionally provide a ``Co-12 cheat sheet'' that acts as a concise summary of our findings on evaluating part-prototype models.

The rising popularity of explainable artificial intelligence (XAI) to understand high-performing black boxes raised the question of how to evaluate explanations of machine learning (ML) models. While interpretability and explainability are often presented as a subjectively validated binary property, we consider it a multi-faceted concept. We identify 12 conceptual properties, such as Compactness and Correctness, that should be evaluated for comprehensively assessing the quality of an explanation. Our so-called Co-12 properties serve as categorization scheme for systematically reviewing the evaluation practices of more than 300 papers published in the past 7 years at major AI and ML conferences that introduce an XAI method. We find that one in three papers evaluate exclusively with anecdotal evidence, and one in five papers evaluate with users. This survey also contributes to the call for objective, quantifiable evaluation methods by presenting an extensive overview of quantitative XAI evaluation methods. Our systematic collection of evaluation methods provides researchers and practitioners with concrete tools to thoroughly validate, benchmark, and compare new and existing XAI methods. The Co-12 categorization scheme and our identified evaluation methods open up opportunities to include quantitative metrics as optimization criteria during model training to optimize for accuracy and interpretability simultaneously.

The rising popularity of explainable artificial intelligence (XAI) to understand high-performing black boxes raised the question of how to evaluate explanations of machine learning (ML) models. While interpretability and explainability are often presented as a subjectively validated binary property, we consider it a multi-faceted concept. We identify 12 conceptual properties, such as Compactness and Correctness, that should be evaluated for comprehensively assessing the quality of an explanation. Our so-called Co-12 properties serve as categorization scheme for systematically reviewing the evaluation practices of more than 300 papers published in the last 7 years at major AI and ML conferences that introduce an XAI method. We find that 1 in 3 papers evaluate exclusively with anecdotal evidence, and 1 in 5 papers evaluate with users. This survey also contributes to the call for objective, quantifiable evaluation methods by presenting an extensive overview of quantitative XAI evaluation methods. Our systematic collection of evaluation methods provides researchers and practitioners with concrete tools to thoroughly validate, benchmark and compare new and existing XAI methods. The Co-12 categorization scheme and our identified evaluation methods open up opportunities to include quantitative metrics as optimization criteria during model training in order to optimize for accuracy and interpretability simultaneously.

A morph is a combination of two separate facial images and contains the identity information of two different people. When used in an identity document, both people can be authenticated by a biometric face recognition (FR) system. Morphs can be generated using either a landmark‐based approach or approaches based on deep learning, such as generative adversarial networks (GANs). In a recent paper, we introduced a worst-case upper bound on how challenging morphing attacks can be for an FR system. The closer morphs are to this upper bound, the bigger the challenge they pose to FR. We introduced an approach with which it was possible to generate morphs that approximate this upper bound for a known FR system (white box) but not for unknown (black box) FR systems. In this paper, we introduce a morph generation method that can approximate worst‐case morphs even when the FR system is not known. A key contribution is that we include the goal of generating difficult morphs during training. Our method is based on adversarially learned inference (ALI) and uses concepts from Wasserstein GANs trained with gradient penalty, which were introduced to stabilise the training of GANs. We include these concepts to achieve a similar improvement in training stability and call the resulting method Wasserstein ALI (WALI). We finetune WALI using loss functions designed specifically to improve the ability to manipulate identity information in facial images and show how it can generate morphs that are more challenging for FR systems than landmark‐ or GAN‐based morphs. We also show how our findings can be used to improve MIPGAN, an existing StyleGAN‐based morph generator.

Machine learning models have been successfully applied for analysis of skin images. However, due to the black box nature of such deep learning models, it is difficult to understand their underlying reasoning. This prevents a human from validating whether the model is right for the right reasons. Spurious correlations and other biases in data can cause a model to base its predictions on such artefacts rather than on the true relevant information. These learned shortcuts can in turn cause incorrect performance estimates and can result in unexpected outcomes when the model is applied in clinical practice. This study presents a method to detect and quantify this shortcut learning in trained classifiers for skin cancer diagnosis, since it is known that dermoscopy images can contain artefacts. Specifically, we train a standard VGG16-based skin cancer classifier on the public ISIC dataset, for which colour calibration charts (elliptical, coloured patches) occur only in benign images and not in malignant ones. Our methodology artificially inserts those patches and uses inpainting to automatically remove patches from images to assess the changes in predictions. We find that our standard classifier partly bases its predictions of benign images on the presence of such a coloured patch. More importantly, by artificially inserting coloured patches into malignant images, we show that shortcut learning results in a significant increase in misdiagnoses, making the classifier unreliable when used in clinical practice. With our results, we, therefore, want to increase awareness of the risks of using black box machine learning models trained on potentially biased datasets. Finally, we present a model-agnostic method to neutralise shortcut learning by removing the bias in the training dataset by exchanging coloured patches with benign skin tissue using image inpainting and re-training the classifier on this de-biased dataset.

Prototype-based methods use interpretable representations to address the black-box nature of deep learning models, in contrast to post-hoc explanation methods that only approximate such models. We propose the Neural Prototype Tree (ProtoTree), an intrinsically interpretable deep learning method for fine-grained image recognition. ProtoTree combines prototype learning with decision trees, and thus results in a globally interpretable model by design. Additionally, ProtoTree can locally explain a single prediction by outlining a decision path through the tree. Each node in our binary tree contains a trainable prototypical part. The presence or absence of this learned prototype in an image determines the routing through a node. Decision making is therefore similar to human reasoning: Does the bird have a red throat? And an elongated beak? Then it's a hummingbird! We tune the accuracy-interpretability trade-off using ensemble methods, pruning and binarizing. We apply pruning without sacrificing accuracy, resulting in a small tree with only 8 learned prototypes along a path to classify a bird from 200 species. An ensemble of 5 ProtoTrees achieves competitive accuracy on the CUB-200- 2011 and Stanford Cars data sets. Code is available at https://github.com/M-Nauta/ProtoTree