Marina M. -C. Höhne

Anna Hedström, Philine Bommer, Kristoffer K. Wickstrøm et al.

One of the unsolved challenges in the field of Explainable AI (XAI) is determining how to most reliably estimate the quality of an explanation method in the absence of ground truth explanation labels. Resolving this issue is of utmost importance as the evaluation outcomes generated by competing evaluation methods (or ''quality estimators''), which aim at measuring the same property of an explanation method, frequently present conflicting rankings. Such disagreements can be challenging for practitioners to interpret, thereby complicating their ability to select the best-performing explanation method. We address this problem through a meta-evaluation of different quality estimators in XAI, which we define as ''the process of evaluating the evaluation method''. Our novel framework, MetaQuantus, analyses two complementary performance characteristics of a quality estimator: its resilience to noise and reactivity to randomness, thus circumventing the need for ground truth labels. We demonstrate the effectiveness of our framework through a series of experiments, targeting various open questions in XAI such as the selection and hyperparameter optimisation of quality estimators. Our work is released under an open-source license (https://github.com/annahedstroem/MetaQuantus) to serve as a development tool for XAI- and Machine Learning (ML) practitioners to verify and benchmark newly constructed quality estimators in a given explainability context. With this work, we provide the community with clear and theoretically-grounded guidance for identifying reliable evaluation methods, thus facilitating reproducibility in the field.

Philine Bommer, Marlene Kretschmer, Anna Hedström et al.

Explainable artificial intelligence (XAI) methods shed light on the predictions of machine learning algorithms. Several different approaches exist and have already been applied in climate science. However, usually missing ground truth explanations complicate their evaluation and comparison, subsequently impeding the choice of the XAI method. Therefore, in this work, we introduce XAI evaluation in the climate context and discuss different desired explanation properties, namely robustness, faithfulness, randomization, complexity, and localization. To this end, we chose previous work as a case study where the decade of annual-mean temperature maps is predicted. After training both a multi-layer perceptron (MLP) and a convolutional neural network (CNN), multiple XAI methods are applied and their skill scores in reference to a random uniform explanation are calculated for each property. Independent of the network, we find that XAI methods Integrated Gradients, layer-wise relevance propagation, and input times gradients exhibit considerable robustness, faithfulness, and complexity while sacrificing randomization performance. Sensitivity methods -- gradient, SmoothGrad, NoiseGrad, and FusionGrad, match the robustness skill but sacrifice faithfulness and complexity for randomization skill. We find architecture-dependent performance differences regarding robustness, complexity and localization skills of different XAI methods, highlighting the necessity for research task-specific evaluation. Overall, our work offers an overview of different evaluation properties in the climate science context and shows how to compare and benchmark different explanation methods, assessing their suitability based on strengths and weaknesses, for the specific research problem at hand. By that, we aim to support climate researchers in the selection of a suitable XAI method.

Xiaoyan Yu, Jannik Franzen, Wojciech Samek et al.

Heatmaps generated on inputs of image classification networks via explainable AI methods like Grad-CAM and LRP have been observed to resemble segmentations of input images in many cases. Consequently, heatmaps have also been leveraged for achieving weakly supervised segmentation with image-level supervision. On the other hand, losses can be imposed on differentiable heatmaps, which has been shown to serve for (1)~improving heatmaps to be more human-interpretable, (2)~regularization of networks towards better generalization, (3)~training diverse ensembles of networks, and (4)~for explicitly ignoring confounding input features. Due to the latter use case, the paradigm of imposing losses on heatmaps is often referred to as "Right for the right reasons". We unify these two lines of research by investigating semi-supervised segmentation as a novel use case for the Right for the Right Reasons paradigm. First, we show formal parallels between differentiable heatmap architectures and standard encoder-decoder architectures for image segmentation. Second, we show that such differentiable heatmap architectures yield competitive results when trained with standard segmentation losses. Third, we show that such architectures allow for training with weak supervision in the form of image-level labels and small numbers of pixel-level labels, outperforming comparable encoder-decoder models. Code is available at \url{https://github.com/Kainmueller-Lab/TW-autoencoder}.

Kirill Bykov, Mayukh Deb, Dennis Grinwald et al.

Deep Neural Networks (DNNs) excel at learning complex abstractions within their internal representations. However, the concepts they learn remain opaque, a problem that becomes particularly acute when models unintentionally learn spurious correlations. In this work, we present DORA (Data-agnOstic Representation Analysis), the first data-agnostic framework for analyzing the representational space of DNNs. Central to our framework is the proposed Extreme-Activation (EA) distance measure, which assesses similarities between representations by analyzing their activation patterns on data points that cause the highest level of activation. As spurious correlations often manifest in features of data that are anomalous to the desired task, such as watermarks or artifacts, we demonstrate that internal representations capable of detecting such artifactual concepts can be found by analyzing relationships within neural representations. We validate the EA metric quantitatively, demonstrating its effectiveness both in controlled scenarios and real-world applications. Finally, we provide practical examples from popular Computer Vision models to illustrate that representations identified as outliers using the EA metric often correspond to undesired and spurious concepts.

Shanghua Liu, Majharulislam Babor, Christoph Verduyn et al.

High resolution phenotyping at the level of individual leaves offers fine-grained insights into plant development and stress responses. However, the full potential of accurate leaf tracking over time remains largely unexplored due to the absence of robust tracking methods-particularly for structurally complex crops such as canola. Existing plant-specific tracking methods are typically limited to small-scale species or rely on constrained imaging conditions. In contrast, generic multi-object tracking (MOT) methods are not designed for dynamic biological scenes. Progress in the development of accurate leaf tracking models has also been hindered by a lack of large-scale datasets captured under realistic conditions. In this work, we introduce CanolaTrack, a new benchmark dataset comprising 5,704 RGB images with 31,840 annotated leaf instances spanning the early growth stages of 184 canola plants. To enable accurate leaf tracking over time, we introduce LeafTrackNet, an efficient framework that combines a YOLOv10-based leaf detector with a MobileNetV3-based embedding network. During inference, leaf identities are maintained over time through an embedding-based memory association strategy. LeafTrackNet outperforms both plant-specific trackers and state-of-the-art MOT baselines, achieving a 9% HOTA improvement on CanolaTrack. With our work we provide a new standard for leaf-level tracking under realistic conditions and we provide CanolaTrack - the largest dataset for leaf tracking in agriculture crops, which will contribute to future research in plant phenotyping. Our code and dataset are publicly available at https://github.com/shl-shawn/LeafTrackNet.

Kirill Bykov, Klaus-Robert Müller, Marina M. -C. Höhne

The utilization of pre-trained networks, especially those trained on ImageNet, has become a common practice in Computer Vision. However, prior research has indicated that a significant number of images in the ImageNet dataset contain watermarks, making pre-trained networks susceptible to learning artifacts such as watermark patterns within their latent spaces. In this paper, we aim to assess the extent to which popular pre-trained architectures display such behavior and to determine which classes are most affected. Additionally, we examine the impact of watermarks on the extracted features. Contrary to the popular belief that the Chinese logographic watermarks impact the "carton" class only, our analysis reveals that a variety of ImageNet classes, such as "monitor", "broom", "apron" and "safe" rely on spurious correlations. Finally, we propose a simple approach to mitigate this issue in fine-tuned networks by ignoring the encodings from the feature-extractor layer of ImageNet pre-trained networks that are most susceptible to watermark imprints.

Kirill Bykov, Laura Kopf, Shinichi Nakajima et al.

Deep Neural Networks (DNNs) demonstrate remarkable capabilities in learning complex hierarchical data representations, but the nature of these representations remains largely unknown. Existing global explainability methods, such as Network Dissection, face limitations such as reliance on segmentation masks, lack of statistical significance testing, and high computational demands. We propose Inverse Recognition (INVERT), a scalable approach for connecting learned representations with human-understandable concepts by leveraging their capacity to discriminate between these concepts. In contrast to prior work, INVERT is capable of handling diverse types of neurons, exhibits less computational complexity, and does not rely on the availability of segmentation masks. Moreover, INVERT provides an interpretable metric assessing the alignment between the representation and its corresponding explanation and delivering a measure of statistical significance. We demonstrate the applicability of INVERT in various scenarios, including the identification of representations affected by spurious correlations, and the interpretation of the hierarchical structure of decision-making within the models.

Pia Hanfeld, Khaled Wahba, Marina M. -C. Höhne et al.

Autonomous flying robots, such as multirotors, often rely on deep learning models that make predictions based on a camera image, e.g. for pose estimation. These models can predict surprising results if applied to input images outside the training domain. This fault can be exploited by adversarial attacks, for example, by computing small images, so-called adversarial patches, that can be placed in the environment to manipulate the neural network's prediction. We introduce flying adversarial patches, where multiple images are mounted on at least one other flying robot and therefore can be placed anywhere in the field of view of a victim multirotor. By introducing the attacker robots, the system is extended to an adversarial multi-robot system. For an effective attack, we compare three methods that simultaneously optimize multiple adversarial patches and their position in the input image. We show that our methods scale well with the number of adversarial patches. Moreover, we demonstrate physical flights with two robots, where we employ a novel attack policy that uses the computed adversarial patches to kidnap a robot that was supposed to follow a human.

Anna Hedström, Leander Weber, Dilyara Bareeva et al.

The evaluation of explanation methods is a research topic that has not yet been explored deeply, however, since explainability is supposed to strengthen trust in artificial intelligence, it is necessary to systematically review and compare explanation methods in order to confirm their correctness. Until now, no tool with focus on XAI evaluation exists that exhaustively and speedily allows researchers to evaluate the performance of explanations of neural network predictions. To increase transparency and reproducibility in the field, we therefore built Quantus -- a comprehensive, evaluation toolkit in Python that includes a growing, well-organised collection of evaluation metrics and tutorials for evaluating explainable methods. The toolkit has been thoroughly tested and is available under an open-source license on PyPi (or on https://github.com/understandable-machine-intelligence-lab/Quantus/).

Dilyara Bareeva, Marina M. -C. Höhne, Alexander Warnecke et al.

Feature Visualization (FV) is a widely used technique for interpreting the concepts learned by Deep Neural Networks (DNNs), which synthesizes input patterns that maximally activate a given feature. Despite its popularity, the trustworthiness of FV explanations has received limited attention. In this paper, we introduce a novel method, Gradient Slingshots, that enables manipulation of FV without modifying the model architecture or significantly degrading its performance. By shaping new trajectories in the off-distribution regions of the activation landscape of a feature, we coerce the optimization process to converge in a predefined visualization. We evaluate our approach on several DNN architectures, demonstrating its ability to replace faithfuls FV with arbitrary targets. These results expose a critical vulnerability: auditors relying solely on FV may accept entirely fabricated explanations. To mitigate this risk, we propose a straightforward defense and quantitatively demonstrate its effectiveness.

Laura Kopf, Nils Feldhus, Kirill Bykov et al.

Automated interpretability research aims to identify concepts encoded in neural network features to enhance human understanding of model behavior. Within the context of large language models (LLMs) for natural language processing (NLP), current automated neuron-level feature description methods face two key challenges: limited robustness and the assumption that each neuron encodes a single concept (monosemanticity), despite increasing evidence of polysemanticity. This assumption restricts the expressiveness of feature descriptions and limits their ability to capture the full range of behaviors encoded in model internals. To address this, we introduce Polysemantic FeatuRe Identification and Scoring Method (PRISM), a novel framework specifically designed to capture the complexity of features in LLMs. Unlike approaches that assign a single description per neuron, common in many automated interpretability methods in NLP, PRISM produces more nuanced descriptions that account for both monosemantic and polysemantic behavior. We apply PRISM to LLMs and, through extensive benchmarking against existing methods, demonstrate that our approach produces more accurate and faithful feature descriptions, improving both overall description quality (via a description score) and the ability to capture distinct concepts when polysemanticity is present (via a polysemanticity score).

Philine L. Bommer, Marlene Kretschmer, Fiona R. Spuler et al.

Predictions on subseasonal-to-seasonal (S2S) timescales--ranging from two weeks to two month--are crucial for early warning systems but remain challenging owing to chaos in the climate system. Teleconnections, such as the stratospheric polar vortex (SPV) and Madden-Julian Oscillation (MJO), offer windows of enhanced predictability, however, their complex interactions remain underutilized in operational forecasting. Here, we developed and evaluated deep learning architectures to predict North Atlantic-European (NAE) weather regimes, systematically assessing the role of remote drivers in improving S2S forecast skill of deep learning models. We implemented (1) a Long Short-term Memory (LSTM) network predicting the NAE regimes of the next six weeks based on previous regimes, (2) an Index-LSTM incorporating SPV and MJO indices, and (3) a ViT-LSTM using a Vision Transformer to directly encode stratospheric wind and tropical outgoing longwave radiation fields. These models are compared with operational hindcasts as well as other AI models. Our results show that leveraging teleconnection information enhances skill at longer lead times. Notably, the ViT-LSTM outperforms ECMWF's subseasonal hindcasts beyond week 4 by improving Scandinavian Blocking (SB) and Atlantic Ridge (AR) predictions. Analysis of high-confidence predictions reveals that NAO-, SB, and AR opportunity forecasts can be associated with SPV variability and MJO phase patterns aligning with established pathways, also indicating new patterns. Overall, our work demonstrates that encoding physically meaningful climate fields can enhance S2S prediction skill, advancing AI-driven subseasonal forecast. Moreover, the experiments highlight the potential of deep learning methods as investigative tools, providing new insights into atmospheric dynamics and predictability.

Carlos Eiras-Franco, Anna Hedström, Marina M. -C. Höhne

Obtaining high-quality explanations of a model's output enables developers to identify and correct biases, align the system's behavior with human values, and ensure ethical compliance. Explainable Artificial Intelligence (XAI) practitioners rely on specific measures to gauge the quality of such explanations. These measures assess key attributes, such as how closely an explanation aligns with a model's decision process (faithfulness), how accurately it pinpoints the relevant input features (localization), and its consistency across different cases (robustness). Despite providing valuable information, these measures do not fully address a critical practitioner's concern: how does the quality of a given explanation compare to other potential explanations? Traditionally, the quality of an explanation has been assessed by comparing it to a randomly generated counterpart. This paper introduces an alternative: the Quality Gap Estimate (QGE). The QGE method offers a direct comparison to what can be viewed as the `inverse' explanation, one that conceptually represents the antithesis of the original explanation. Our extensive testing across multiple model architectures, datasets, and established quality metrics demonstrates that the QGE method is superior to the traditional approach. Furthermore, we show that QGE enhances the statistical reliability of these quality assessments. This advance represents a significant step toward a more insightful evaluation of explanations that enables a more effective inspection of a model's behavior.

Shanghua Liu, Anna Hedström, Deepak Hanike Basavegowda et al.

Grasslands are known for their high biodiversity and ability to provide multiple ecosystem services. Challenges in automating the identification of indicator plants are key obstacles to large-scale grassland monitoring. These challenges stem from the scarcity of extensive datasets, the distributional shifts between generic and grassland-specific datasets, and the inherent opacity of deep learning models. This paper delves into the latter two challenges, with a specific focus on transfer learning and eXplainable Artificial Intelligence (XAI) approaches to grassland monitoring, highlighting the novelty of XAI in this domain. We analyze various transfer learning methods to bridge the distributional gaps between generic and grassland-specific datasets. Additionally, we showcase how explainable AI techniques can unveil the model's domain adaptation capabilities, employing quantitative assessments to evaluate the model's proficiency in accurately centering relevant input features around the object of interest. This research contributes valuable insights for enhancing model performance through transfer learning and measuring domain adaptability with explainable AI, showing significant promise for broader applications within the agricultural community.

Pia Hanfeld, Marina M. -C. Höhne, Michael Bussmann et al.

Autonomous flying robots, e.g. multirotors, often rely on a neural network that makes predictions based on a camera image. These deep learning (DL) models can compute surprising results if applied to input images outside the training domain. Adversarial attacks exploit this fault, for example, by computing small images, so-called adversarial patches, that can be placed in the environment to manipulate the neural network's prediction. We introduce flying adversarial patches, where an image is mounted on another flying robot and therefore can be placed anywhere in the field of view of a victim multirotor. For an effective attack, we compare three methods that simultaneously optimize the adversarial patch and its position in the input image. We perform an empirical validation on a publicly available DL model and dataset for autonomous multirotors. Ultimately, our attacking multirotor would be able to gain full control over the motions of the victim multirotor.

Dennis Grinwald, Kirill Bykov, Shinichi Nakajima et al.

Explainable Artificial Intelligence (XAI) aims to make learning machines less opaque, and offers researchers and practitioners various tools to reveal the decision-making strategies of neural networks. In this work, we investigate how XAI methods can be used for exploring and visualizing the diversity of feature representations learned by Bayesian Neural Networks (BNNs). Our goal is to provide a global understanding of BNNs by making their decision-making strategies a) visible and tangible through feature visualizations and b) quantitatively measurable with a distance measure learned by contrastive learning. Our work provides new insights into the \emph{posterior} distribution in terms of human-understandable feature information with regard to the underlying decision making strategies. The main findings of our work are the following: 1) global XAI methods can be applied to explain the diversity of decision-making strategies of BNN instances, 2) Monte Carlo dropout with commonly used Dropout rates exhibit increased diversity in feature representations compared to the multimodal posterior approximation of MultiSWAG, 3) the diversity of learned feature representations highly correlates with the uncertainty estimate for the output and 4) the inter-mode diversity of the multimodal posterior decreases as the network width increases, while the intra mode diversity increases. These findings are consistent with the recent Deep Neural Networks theory, providing additional intuitions about what the theory implies in terms of humanly understandable concepts.

Srishti Gautam, Marina M. -C. Höhne, Stine Hansen et al.

The recent trend of integrating multi-source Chest X-Ray datasets to improve automated diagnostics raises concerns that models learn to exploit source-specific correlations to improve performance by recognizing the source domain of an image rather than the medical pathology. We hypothesize that this effect is enforced by and leverages label-imbalance across the source domains, i.e, prevalence of a disease corresponding to a source. Therefore, in this work, we perform a thorough study of the effect of label-imbalance in multi-source training for the task of pneumonia detection on the widely used ChestX-ray14 and CheXpert datasets. The results highlight and stress the importance of using more faithful and transparent self-explaining models for automated diagnosis, thus enabling the inherent detection of spurious learning. They further illustrate that this undesirable effect of learning spurious correlations can be reduced considerably when ensuring label-balanced source domain datasets.

Yamen Ali, Aiham Taleb, Marina M. -C. Höhne et al.

Self-supervised learning methods can be used to learn meaningful representations from unlabeled data that can be transferred to supervised downstream tasks to reduce the need for labeled data. In this paper, we propose a 3D self-supervised method that is based on the contrastive (SimCLR) method. Additionally, we show that employing Bayesian neural networks (with Monte-Carlo Dropout) during the inference phase can further enhance the results on the downstream tasks. We showcase our models on two medical imaging segmentation tasks: i) Brain Tumor Segmentation from 3D MRI, ii) Pancreas Tumor Segmentation from 3D CT. Our experimental results demonstrate the benefits of our proposed methods in both downstream data-efficiency and performance.

Srishti Gautam, Marina M. -C. Höhne, Stine Hansen et al.

Current machine learning models have shown high efficiency in solving a wide variety of real-world problems. However, their black box character poses a major challenge for the understanding and traceability of the underlying decision-making strategies. As a remedy, many post-hoc explanation and self-explanatory methods have been developed to interpret the models' behavior. These methods, in addition, enable the identification of artifacts that can be learned by the model as class-relevant features. In this work, we provide a detailed case study of the self-explaining network, ProtoPNet, in the presence of a spectrum of artifacts. Accordingly, we identify the main drawbacks of ProtoPNet, especially, its coarse and spatially imprecise explanations. We address these limitations by introducing Prototypical Relevance Propagation (PRP), a novel method for generating more precise model-aware explanations. Furthermore, in order to obtain a clean dataset, we propose to use multi-view clustering strategies for segregating the artifact images using the PRP explanations, thereby suppressing the potential artifact learning in the models.

Kirill Bykov, Marina M. -C. Höhne, Adelaida Creosteanu et al.

To advance the transparency of learning machines such as Deep Neural Networks (DNNs), the field of Explainable AI (XAI) was established to provide interpretations of DNNs' predictions. While different explanation techniques exist, a popular approach is given in the form of attribution maps, which illustrate, given a particular data point, the relevant patterns the model has used for making its prediction. Although Bayesian models such as Bayesian Neural Networks (BNNs) have a limited form of transparency built-in through their prior weight distribution, they lack explanations of their predictions for given instances. In this work, we take a step toward combining these two perspectives by examining how local attributions can be extended to BNNs. Within the Bayesian framework, network weights follow a probability distribution; hence, the standard point explanation extends naturally to an explanation distribution. Viewing explanations probabilistically, we aggregate and analyze multiple local attributions drawn from an approximate posterior to explore variability in explanation patterns. The diversity of explanations offers a way to further explore how predictive rationales may vary across posterior samples. Quantitative and qualitative experiments on toy and benchmark data, as well as on a real-world pathology dataset, illustrate that our framework enriches standard explanations with uncertainty information and may support the visualization of explanation stability.

Kirill Bykov, Anna Hedström, Shinichi Nakajima et al.

Many efforts have been made for revealing the decision-making process of black-box learning machines such as deep neural networks, resulting in useful local and global explanation methods. For local explanation, stochasticity is known to help: a simple method, called SmoothGrad, has improved the visual quality of gradient-based attribution by adding noise to the input space and averaging the explanations of the noisy inputs. In this paper, we extend this idea and propose NoiseGrad that enhances both local and global explanation methods. Specifically, NoiseGrad introduces stochasticity in the weight parameter space, such that the decision boundary is perturbed. NoiseGrad is expected to enhance the local explanation, similarly to SmoothGrad, due to the dual relationship between the input perturbation and the decision boundary perturbation. We evaluate NoiseGrad and its fusion with SmoothGrad -- FusionGrad -- qualitatively and quantitatively with several evaluation criteria, and show that our novel approach significantly outperforms the baseline methods. Both NoiseGrad and FusionGrad are method-agnostic and as handy as SmoothGrad using a simple heuristic for the choice of the hyperparameter setting without the need of finetuning.

Kirill Bykov, Marina M. -C. Höhne, Klaus-Robert Müller et al.

Explainable AI (XAI) aims to provide interpretations for predictions made by learning machines, such as deep neural networks, in order to make the machines more transparent for the user and furthermore trustworthy also for applications in e.g. safety-critical areas. So far, however, no methods for quantifying uncertainties of explanations have been conceived, which is problematic in domains where a high confidence in explanations is a prerequisite. We therefore contribute by proposing a new framework that allows to convert any arbitrary explanation method for neural networks into an explanation method for Bayesian neural networks, with an in-built modeling of uncertainties. Within the Bayesian framework a network's weights follow a distribution that extends standard single explanation scores and heatmaps to distributions thereof, in this manner translating the intrinsic network model uncertainties into a quantification of explanation uncertainties. This allows us for the first time to carve out uncertainties associated with a model explanation and subsequently gauge the appropriate level of explanation confidence for a user (using percentiles). We demonstrate the effectiveness and usefulness of our approach extensively in various experiments, both qualitatively and quantitatively.