Case Study: Ensemble Decision-Based Annotation of Unconstrained Real Estate Images

We describe a proof-of-concept for annotating real estate images using simple iterative rule-based semi-supervised learning. In this study, we have gained important insights into the content characteristics and uniqueness of individual image classes as well as essential requirements for a practical implementation.

Explaining YOLO: Leveraging Grad-CAM to Explain Object Detections

We investigate the problem of explainability for visual object detectors. Specifically, we demonstrate on the example of the YOLO object detector how to integrate Grad-CAM into the model architecture and analyze the results. We show how to compute attribution-based explanations for individual detections and find that the normalization of the results has a great impact on their interpretation.

Real Estate Attribute Prediction from Multiple Visual Modalities with Missing Data

The assessment and valuation of real estate requires large datasets with real estate information. Unfortunately, real estate databases are usually sparse in practice, i.e., not for each property every important attribute is available. In this paper, we study the potential of predicting high-level real estate attributes from visual data, specifically from two visual modalities, namely indoor (interior) and outdoor (facade) photos. We design three models using different multimodal fusion strategies and evaluate them for three different use cases. Thereby, a particular challenge is to handle missing modalities. We evaluate different fusion strategies, present baselines for the different prediction tasks, and find that enriching the training data with additional incomplete samples can lead to an improvement in prediction accuracy. Furthermore, the fusion of information from indoor and outdoor photos results in a performance boost of up to 5% in Macro F1-score.

Trustworthy Visual Analytics in Clinical Gait Analysis: A Case Study for Patients with Cerebral Palsy

Three-dimensional clinical gait analysis is essential for selecting optimal treatment interventions for patients with cerebral palsy (CP), but generates a large amount of time series data. For the automated analysis of these data, machine learning approaches yield promising results. However, due to their black-box nature, such approaches are often mistrusted by clinicians. We propose gaitXplorer, a visual analytics approach for the classification of CP-related gait patterns that integrates Grad-CAM, a well-established explainable artificial intelligence algorithm, for explanations of machine learning classifications. Regions of high relevance for classification are highlighted in the interactive visual interface. The approach is evaluated in a case study with two clinical gait experts. They inspected the explanations for a sample of eight patients using the visual interface and expressed which relevance scores they found trustworthy and which they found suspicious. Overall, the clinicians gave positive feedback on the approach as it allowed them a better understanding of which regions in the data were relevant for the classification.

Multimodal Detection of Information Disorder from Social Media

Social media is accompanied by an increasing proportion of content that provides fake information or misleading content, known as information disorder. In this paper, we study the problem of multimodal fake news detection on a largescale multimodal dataset. We propose a multimodal network architecture that enables different levels and types of information fusion. In addition to the textual and visual content of a posting, we further leverage secondary information, i.e. user comments and metadata. We fuse information at multiple levels to account for the specific intrinsic structure of the modalities. Our results show that multimodal analysis is highly effective for the task and all modalities contribute positively when fused properly.

Bounded logit attention: Learning to explain image classifiers

Explainable artificial intelligence is the attempt to elucidate the workings of systems too complex to be directly accessible to human cognition through suitable side-information referred to as "explanations". We present a trainable explanation module for convolutional image classifiers we call bounded logit attention (BLA). The BLA module learns to select a subset of the convolutional feature map for each input instance, which then serves as an explanation for the classifier's prediction. BLA overcomes several limitations of the instancewise feature selection method "learning to explain" (L2X) introduced by Chen et al. (2018): 1) BLA scales to real-world sized image classification problems, and 2) BLA offers a canonical way to learn explanations of variable size. Due to its modularity BLA lends itself to transfer learning setups and can also be employed as a post-hoc add-on to trained classifiers. Beyond explainability, BLA may serve as a general purpose method for differentiable approximation of subset selection. In a user study we find that BLA explanations are preferred over explanations generated by the popular (Grad-)CAM method.

$k$-Anonymity in Practice: How Generalisation and Suppression Affect Machine Learning Classifiers

The protection of private information is a crucial issue in data-driven research and business contexts. Typically, techniques like anonymisation or (selective) deletion are introduced in order to allow data sharing, \eg\ in the case of collaborative research endeavours. For use with anonymisation techniques, the $k$-anonymity criterion is one of the most popular, with numerous scientific publications on different algorithms and metrics. Anonymisation techniques often require changing the data and thus necessarily affect the results of machine learning models trained on the underlying data. In this work, we conduct a systematic comparison and detailed investigation into the effects of different $k$-anonymisation algorithms on the results of machine learning models. We investigate a set of popular $k$-anonymisation algorithms with different classifiers and evaluate them on different real-world datasets. Our systematic evaluation shows that with an increasingly strong $k$-anonymity constraint, the classification performance generally degrades, but to varying degrees and strongly depending on the dataset and anonymisation method. Furthermore, Mondrian can be considered as the method with the most appealing properties for subsequent classification.

Machine Unlearning: Linear Filtration for Logit-based Classifiers

Recently enacted legislation grants individuals certain rights to decide in what fashion their personal data may be used, and in particular a "right to be forgotten". This poses a challenge to machine learning: how to proceed when an individual retracts permission to use data which has been part of the training process of a model? From this question emerges the field of machine unlearning, which could be broadly described as the investigation of how to "delete training data from models". Our work complements this direction of research for the specific setting of class-wide deletion requests for classification models (e.g. deep neural networks). As a first step, we propose linear filtration as a computationally efficient sanitization method. Our experiments demonstrate benefits in an adversarial setting over naive deletion schemes.

On the Understanding and Interpretation of Machine Learning Predictions in Clinical Gait Analysis Using Explainable Artificial Intelligence

Systems incorporating Artificial Intelligence (AI) and machine learning (ML) techniques are increasingly used to guide decision-making in the healthcare sector. While AI-based systems provide powerful and promising results with regard to their classification and prediction accuracy (e.g., in differentiating between different disorders in human gait), most share a central limitation, namely their black-box character. Understanding which features classification models learn, whether they are meaningful and consequently whether their decisions are trustworthy is difficult and often impossible to comprehend. This severely hampers their applicability as decision-support systems in clinical practice. There is a strong need for AI-based systems to provide transparency and justification of predictions, which are necessary also for ethical and legal compliance. As a consequence, in recent years the field of explainable AI (XAI) has gained increasing importance. The primary aim of this article is to investigate whether XAI methods can enhance transparency, explainability and interpretability of predictions in automated clinical gait classification. We utilize a dataset comprising bilateral three-dimensional ground reaction force measurements from 132 patients with different lower-body gait disorders and 62 healthy controls. In our experiments, we included several gait classification tasks, employed a representative set of classification methods, and a well-established XAI method - Layer-wise Relevance Propagation - to explain decisions at the signal (input) level. The presented approach exemplifies how XAI can be used to understand and interpret state-of-the-art ML models trained for gait classification tasks, and shows that the features that are considered relevant for machine learning models can be attributed to meaningful and clinically relevant biomechanical gait characteristics.

Persistence Bag-of-Words for Topological Data Analysis

Persistent homology (PH) is a rigorous mathematical theory that provides a robust descriptor of data in the form of persistence diagrams (PDs). PDs are compact 2D representations formed by multisets of points. Their variable size makes them, however, difficult to combine with typical machine learning workflows. In this paper, we introduce persistence bag-of-words, which is a novel, expressive and discriminative vectorized representation of PDs for topological data analysis. It represents PDs in a convenient way for machine learning and statistical analysis and has a number of favorable practical and theoretical properties like 1-Wasserstein stability. We evaluate our representation on several heterogeneous datasets and show its high discriminative power. Our approach achieves state-of-the-art performance and even beyond in much less time than alternative approaches. Thereby, it facilitates the topological analysis of large-scale data sets in future.