Research Papers

We present Azimuth, an open-source and easy-to-use tool to perform error analysis for text classification. Compared to other stages of the ML development cycle, such as model training and hyper-parameter tuning, the process and tooling for the error analysis stage are less mature. However, this stage is critical for the development of reliable and trustworthy AI systems. To make error analysis more systematic, we propose an approach comprising dataset analysis and model quality assessment, which Azimuth facilitates. We aim to help AI practitioners discover and address areas where the model does not generalize by leveraging and integrating a range of ML techniques, such as saliency maps, similarity, uncertainty, and behavioral analyses, all in one tool. Our code and documentation are available at github.com/servicenow/azimuth.

Towards Human-Interpretable Prototypes for Visual Assessment of Image Classification Models

Poulami Sinhamahapatra, Lena Heidemann, Maureen Monnet +1

Explaining black-box Artificial Intelligence (AI) models is a cornerstone for trustworthy AI and a prerequisite for its use in safety critical applications such that AI models can reliably assist humans in critical decisions. However, instead of trying to explain our models post-hoc, we need models which are interpretable-by-design built on a reasoning process similar to humans that exploits meaningful high-level concepts such as shapes, texture or object parts. Learning such concepts is often hindered by its need for explicit specification and annotation up front. Instead, prototype-based learning approaches such as ProtoPNet claim to discover visually meaningful prototypes in an unsupervised way. In this work, we propose a set of properties that those prototypes have to fulfill to enable human analysis, e.g. as part of a reliable model assessment case, and analyse such existing methods in the light of these properties. Given a 'Guess who?' game, we find that these prototypes still have a long way ahead towards definite explanations. We quantitatively validate our findings by conducting a user study indicating that many of the learnt prototypes are not considered useful towards human understanding. We discuss about the missing links in the existing methods and present a potential real-world application motivating the need to progress towards truly human-interpretable prototypes.

Explainable artificial intelligence is an emerging and evolving concept. Its impact on construction, though yet to be realised, will be profound in the foreseeable future. Still, XAI has received limited attention in construction. As a result, no evaluation frameworks have been propagated to enable construction organisations to understand the what, why, how, and when of XAI. Our paper aims to fill this void by developing a content, context, process, and outcome evaluation framework that can be used to justify the adoption and effective management of XAI. After introducing and describing this novel framework, we discuss its implications for future research. While our novel framework is conceptual, it provides a frame of reference for construction organisations to make headway toward realising XAI business value and benefits.

Evaluating Search System Explainability with Psychometrics and Crowdsourcing

Catherine Chen, Carsten Eickhoff

As information retrieval (IR) systems, such as search engines and conversational agents, become ubiquitous in various domains, the need for transparent and explainable systems grows to ensure accountability, fairness, and unbiased results. Despite recent advances in explainable AI and IR techniques, there is no consensus on the definition of explainability. Existing approaches often treat it as a singular notion, disregarding the multidimensional definition postulated in the literature. In this paper, we use psychometrics and crowdsourcing to identify human-centered factors of explainability in Web search systems and introduce SSE (Search System Explainability), an evaluation metric for explainable IR (XIR) search systems. In a crowdsourced user study, we demonstrate SSE's ability to distinguish between explainable and non-explainable systems, showing that systems with higher scores indeed indicate greater interpretability. We hope that aside from these concrete contributions to XIR, this line of work will serve as a blueprint for similar explainability evaluation efforts in other domains of machine learning and natural language processing.

Banks hold a societal responsibility and regulatory requirements to mitigate the risk of financial crimes. Risk mitigation primarily happens through monitoring customer activity through Transaction Monitoring (TM). Recently, Machine Learning (ML) has been proposed to identify suspicious customer behavior, which raises complex socio-technical implications around trust and explainability of ML models and their outputs. However, little research is available due to its sensitivity. We aim to fill this gap by presenting empirical research exploring how ML supported automation and augmentation affects the TM process and stakeholders' requirements for building eXplainable Artificial Intelligence (xAI). Our study finds that xAI requirements depend on the liable party in the TM process which changes depending on augmentation or automation of TM. Context-relatable explanations can provide much-needed support for auditing and may diminish bias in the investigator's judgement. These results suggest a use case-specific approach for xAI to adequately foster the adoption of ML in TM.

Intelligent systems that use Machine Learning classification algorithms are increasingly common in everyday society. However, many systems use black-box models that do not have characteristics that allow for self-explanation of their predictions. This situation leads researchers in the field and society to the following question: How can I trust the prediction of a model I cannot understand? In this sense, XAI emerges as a field of AI that aims to create techniques capable of explaining the decisions of the classifier to the end-user. As a result, several techniques have emerged, such as Explanation-by-Example, which has a few initiatives consolidated by the community currently working with XAI. This research explores the Item Response Theory (IRT) as a tool to explaining the models and measuring the level of reliability of the Explanation-by-Example approach. To this end, four datasets with different levels of complexity were used, and the Random Forest model was used as a hypothesis test. From the test set, 83.8% of the errors are from instances in which the IRT points out the model as unreliable.

Transcending XAI Algorithm Boundaries through End-User-Inspired Design

Weina Jin, Jianyu Fan, Diane Gromala +3

The boundaries of existing explainable artificial intelligence (XAI) algorithms are confined to problems grounded in technical users' demand for explainability. This research paradigm disproportionately ignores the larger group of non-technical end users, who have a much higher demand for AI explanations in diverse explanation goals, such as making safer and better decisions and improving users' predicted outcomes. Lacking explainability-focused functional support for end users may hinder the safe and accountable use of AI in high-stakes domains, such as healthcare, criminal justice, finance, and autonomous driving systems. Built upon prior human factor analysis on end users' requirements for XAI, we identify and model four novel XAI technical problems covering the full spectrum from design to the evaluation of XAI algorithms, including edge-case-based reasoning, customizable counterfactual explanation, collapsible decision tree, and the verifiability metric to evaluate XAI utility. Based on these newly-identified research problems, we also discuss open problems in the technical development of user-centered XAI to inspire future research. Our work bridges human-centered XAI with the technical XAI community, and calls for a new research paradigm on the technical development of user-centered XAI for the responsible use of AI in critical tasks.

Explainable artificial intelligence (XAI) is increasingly used to analyze the behavior of neural networks. Concept activation uses human-interpretable concepts to explain neural network behavior. This study aimed at assessing the feasibility of regression concept activation to explain detection and classification of multi-modal volumetric data. Proof-of-concept was demonstrated in metastatic prostate cancer patients imaged with positron emission tomography/computed tomography (PET/CT). Multi-modal volumetric concept activation was used to provide global and local explanations. Sensitivity was 80% at 1.78 false positive per patient. Global explanations showed that detection focused on CT for anatomical location and on PET for its confidence in the detection. Local explanations showed promise to aid in distinguishing true positives from false positives. Hence, this study demonstrated feasibility to explain detection and classification of multi-modal volumetric data using regression concept activation.

Explaining Causal Models with Argumentation: the Case of Bi-variate Reinforcement

Antonio Rago, Pietro Baroni, Francesca Toni

Causal models are playing an increasingly important role in machine learning, particularly in the realm of explainable AI. We introduce a conceptualisation for generating argumentation frameworks (AFs) from causal models for the purpose of forging explanations for the models' outputs. The conceptualisation is based on reinterpreting desirable properties of semantics of AFs as explanation moulds, which are means for characterising the relations in the causal model argumentatively. We demonstrate our methodology by reinterpreting the property of bi-variate reinforcement as an explanation mould to forge bipolar AFs as explanations for the outputs of causal models. We perform a theoretical evaluation of these argumentative explanations, examining whether they satisfy a range of desirable explanatory and argumentative properties.

Scientific Explanation and Natural Language: A Unified Epistemological-Linguistic Perspective for Explainable AI

Marco Valentino, André Freitas

A fundamental research goal for Explainable AI (XAI) is to build models that are capable of reasoning through the generation of natural language explanations. However, the methodologies to design and evaluate explanation-based inference models are still poorly informed by theoretical accounts on the nature of explanation. As an attempt to provide an epistemologically grounded characterisation for XAI, this paper focuses on the scientific domain, aiming to bridge the gap between theory and practice on the notion of a scientific explanation. Specifically, the paper combines a detailed survey of the modern accounts of scientific explanation in Philosophy of Science with a systematic analysis of corpora of natural language explanations, clarifying the nature and function of explanatory arguments from both a top-down (categorical) and a bottom-up (corpus-based) perspective. Through a mixture of quantitative and qualitative methodologies, the presented study allows deriving the following main conclusions: (1) Explanations cannot be entirely characterised in terms of inductive or deductive arguments as their main function is to perform unification; (2) An explanation must cite causes and mechanisms that are responsible for the occurrence of the event to be explained; (3) While natural language explanations possess an intrinsic causal-mechanistic nature, they are not limited to causes and mechanisms, also accounting for pragmatic elements such as definitions, properties and taxonomic relations; (4) Patterns of unification naturally emerge in corpora of explanations even if not intentionally modelled; (5) Unification is realised through a process of abstraction, whose function is to provide the inference substrate for subsuming the event to be explained under recurring patterns and high-level regularities.

Recent works have recognized the need for human-centered perspectives when designing and evaluating human-AI interactions and explainable AI methods. Yet, current approaches fall short at intercepting and managing unexpected user behavior resulting from the interaction with AI systems and explainability methods of different stake-holder groups. In this work, we explore the use of AI and explainability methods in the insurance domain. In an qualitative case study with participants with different roles and professional backgrounds, we show that AI and explainability methods are used in creative ways in daily workflows, resulting in a divergence between their intended and actual use. Finally, we discuss some recommendations for the design of human-AI interactions and explainable AI methods to manage the risks and harness the potential of unexpected user behavior.

Predicitions made by neural networks can be fraudulently altered by so-called poisoning attacks. A special case are backdoor poisoning attacks. We study suitable detection methods and introduce a new method called Heatmap Clustering. There, we apply a $k$-means clustering algorithm on heatmaps produced by the state-of-the-art explainable AI method Layer-wise relevance propagation. The goal is to separate poisoned from un-poisoned data in the dataset. We compare this method with a similar method, called Activation Clustering, which also uses $k$-means clustering but applies it on the activation of certain hidden layers of the neural network as input. We test the performance of both approaches for standard backdoor poisoning attacks, label-consistent poisoning attacks and label-consistent poisoning attacks with reduced amplitude stickers. We show that Heatmap Clustering consistently performs better than Activation Clustering. However, when considering label-consistent poisoning attacks, the latter method also yields good detection performance.