Research Papers

Unexplainable black-box models create scenarios where anomalies cause deleterious responses, thus creating unacceptable risks. These risks have motivated the field of eXplainable Artificial Intelligence (XAI) to improve trust by evaluating local interpretability in black-box neural networks. Unfortunately, the ground truth is unavailable for the model's decision, so evaluation is limited to qualitative assessment. Further, interpretability may lead to inaccurate conclusions about the model or a false sense of trust. We propose to improve XAI from the vantage point of the user's trust by exploring a black-box model's latent feature space. We present an approach, ProtoShotXAI, that uses a Prototypical few-shot network to explore the contrastive manifold between nonlinear features of different classes. A user explores the manifold by perturbing the input features of a query sample and recording the response for a subset of exemplars from any class. Our approach is the first locally interpretable XAI model that can be extended to, and demonstrated on, few-shot networks. We compare ProtoShotXAI to the state-of-the-art XAI approaches on MNIST, Omniglot, and ImageNet to demonstrate, both quantitatively and qualitatively, that ProtoShotXAI provides more flexibility for model exploration. Finally, ProtoShotXAI also demonstrates novel explainabilty and detectabilty on adversarial samples.

Deep learning has become popular because of its potential to achieve high accuracy in prediction tasks. However, accuracy is not always the only goal of statistical modelling, especially for models developed as part of scientific research. Rather, many scientific models are developed to facilitate scientific discovery, by which we mean to abstract a human-understandable representation of the natural world. Unfortunately, the opacity of deep neural networks limit their role in scientific discovery, creating a new demand for models that are transparently interpretable. This article is a field guide to transparent model design. It provides a taxonomy of transparent model design concepts, a practical workflow for putting design concepts into practice, and a general template for reporting design choices. We hope this field guide will help researchers more effectively design transparently interpretable models, and thus enable them to use deep learning for scientific discovery.

There is a struggle in Artificial intelligence (AI) ethics to gain ground in actionable methods and models to be utilized by practitioners while developing and implementing ethically sound AI systems. AI ethics is a vague concept without a consensus of definition or theoretical grounding and bearing little connection to practice. Practice involving primarily technical tasks like software development is not aptly equipped to process and decide upon ethical considerations. Efforts to create tools and guidelines to help people working with AI development have been concentrating almost solely on the technical aspects of AI. A few exceptions do apply, such as the ECCOLA method for creating ethically aligned AI -systems. ECCOLA has proven results in terms of increased ethical considerations in AI systems development. Yet, it is a novel innovation, and room for development still exists. This study aims to extend ECCOLA with a deployment model to drive the adoption of ECCOLA, as any method, no matter how good, is of no value without adoption and use. The model includes simple metrics to facilitate the communication of ethical gaps or outcomes of ethical AI development. It offers the opportunity to assess any AI system at any given lifecycle phase, e.g., opening possibilities like analyzing the ethicality of an AI system under acquisition.

Algorithmic bias is of increasing concern, both to the research community, and society at large. Bias in AI is more abstract and unintuitive than traditional forms of discrimination and can be more difficult to detect and mitigate. A clear gap exists in the current literature on evaluating the relative bias in the performance of multi-class classifiers. In this work, we propose two simple yet effective metrics, Combined Error Variance (CEV) and Symmetric Distance Error (SDE), to quantitatively evaluate the class-wise bias of two models in comparison to one another. By evaluating the performance of these new metrics and by demonstrating their practical application, we show that they can be used to measure fairness as well as bias. These demonstrations show that our metrics can address specific needs for measuring bias in multi-class classification.

This position paper highlights an ethic of machine-to-machine cooperation and machine pro-sociality, and argues that machines capable of autonomous sensing, decision-making and action, such as automated vehicles and urban robots, owned and used by different self-interested parties, and having their own agendas (or interests of their owners) should be designed and built to be cooperative in their behaviours, especially if they share public spaces. That is, by design, the machine should first cooperate, and then only consider alternatives if there are problems. It is argued that being cooperative is not only important for their improved functioning, especially, when they use shared resources (e.g., parking spaces, public roads, curbside space and walkways), but also as a favourable requirement analogous to how humans cooperating with other humans can be advantageous and often viewed favourably. The usefulness of such machine-to-machine cooperation are illustrated via examples including cooperative crowdsourcing, cooperative traffic routing and parking as well as futuristic scenarios involving urban robots for delivery and shopping. It is argued that just as privacy-by-design and security-by-design are important considerations, in order to yield systems that fulfil ethical requirements, cooperative-by-design should also be an imperative for autonomous systems that are separately owned but co-inhabit the same spaces and use common resources. If a machine using shared public spaces is not cooperative, as one might expect, then it is not only anti-social but not behaving ethically. It is also proposed that certification for urban robots that operate in public could be explored.

When Stability meets Sufficiency: Informative Explanations that do not Overwhelm

Ronny Luss, Amit Dhurandhar

Recent studies evaluating various criteria for explainable artificial intelligence (XAI) suggest that fidelity, stability, and comprehensibility are among the most important metrics considered by users of AI across a diverse collection of usage contexts. We consider these criteria as applied to feature-based attribution methods, which are amongst the most prevalent in XAI literature. Going beyond standard correlation, methods have been proposed that highlight what should be minimally sufficient to justify the classification of an input (viz. pertinent positives). While minimal sufficiency is an attractive property akin to comprehensibility, the resulting explanations are often too sparse for a human to understand and evaluate the local behavior of the model. To overcome these limitations, we incorporate the criteria of stability and fidelity and propose a novel method called Path-Sufficient Explanations Method (PSEM) that outputs a sequence of stable and sufficient explanations for a given input of strictly decreasing size (or value) -- from original input to a minimally sufficient explanation -- which can be thought to trace the local boundary of the model in a stable manner, thus providing better intuition about the local model behavior for the specific input. We validate these claims, both qualitatively and quantitatively, with experiments that show the benefit of PSEM across three modalities (image, tabular and text) as well as versus other path explanations. A user study depicts the strength of the method in communicating the local behavior, where (many) users are able to correctly determine the prediction made by a model.

Traditional adversarial examples are typically generated by adding perturbation noise to the input image within a small matrix norm. In practice, un-restricted adversarial attack has raised great concern and presented a new threat to the AI safety. In this paper, we propose a wavelet-VAE structure to reconstruct an input image and generate adversarial examples by modifying the latent code. Different from perturbation-based attack, the modifications of the proposed method are not limited but imperceptible to human eyes. Experiments show that our method can generate high quality adversarial examples on ImageNet dataset.

Thanks to the great progress of machine learning in the last years, several Artificial Intelligence (AI) techniques have been increasingly moving from the controlled research laboratory settings to our everyday life. AI is clearly supportive in many decision-making scenarios, but when it comes to sensitive areas such as health care, hiring policies, education, banking or justice, with major impact on individuals and society, it becomes crucial to establish guidelines on how to design, develop, deploy and monitor this technology. Indeed the decision rules elaborated by machine learning models are data-driven and there are multiple ways in which discriminatory biases can seep into data. Algorithms trained on those data incur the risk of amplifying prejudices and societal stereotypes by over associating protected attributes such as gender, ethnicity or disabilities with the prediction task. Starting from the extensive experience of the National Metrology Institute on measurement standards and certification roadmaps, and of Politecnico di Torino on machine learning as well as methods for domain bias evaluation and mastering, we propose a first joint effort to define the operational steps needed for AI fairness certification. Specifically we will overview the criteria that should be met by an AI system before coming into official service and the conformity assessment procedures useful to monitor its functioning for fair decisions.

Be Considerate: Objectives, Side Effects, and Deciding How to Act

Parand Alizadeh Alamdari, Toryn Q. Klassen, Rodrigo Toro Icarte +1

Recent work in AI safety has highlighted that in sequential decision making, objectives are often underspecified or incomplete. This gives discretion to the acting agent to realize the stated objective in ways that may result in undesirable outcomes. We contend that to learn to act safely, a reinforcement learning (RL) agent should include contemplation of the impact of its actions on the wellbeing and agency of others in the environment, including other acting agents and reactive processes. We endow RL agents with the ability to contemplate such impact by augmenting their reward based on expectation of future return by others in the environment, providing different criteria for characterizing impact. We further endow these agents with the ability to differentially factor this impact into their decision making, manifesting behavior that ranges from self-centred to self-less, as demonstrated by experiments in gridworld environments.

Artificial intelligence (AI) is becoming increasingly more popular and can be found in workplaces and homes around the world. The decisions made by such "black box" systems are often opaque; that is, so complex as to be functionally impossible to understand. How do we ensure that these systems are behaving as desired? TrustyAI is an initiative which looks into explainable artificial intelligence (XAI) solutions to address this issue of explainability in the context of both AI models and decision services. This paper presents the TrustyAI Explainability Toolkit, a Java and Python library that provides XAI explanations of decision services and predictive models for both enterprise and data science use-cases. We describe the TrustyAI implementations and extensions to techniques such as LIME, SHAP and counterfactuals, which are benchmarked against existing implementations in a variety of experiments.

Explainable AI (XAI) methods focus on explaining what a neural network has learned - in other words, identifying the features that are the most influential to the prediction. In this paper, we call them "distinguishing features". However, whether a human can make sense of the generated explanation also depends on the perceptibility of these features to humans. To make sure an explanation is human-understandable, we argue that the capabilities of humans, constrained by the Human Visual System (HVS) and psychophysics, need to be taken into account. We propose the {\em human perceptibility principle for XAI}, stating that, to generate human-understandable explanations, neural networks should be steered towards focusing on human-understandable cues during training. We conduct a case study regarding the classification of real vs. fake face images, where many of the distinguishing features picked up by standard neural networks turn out not to be perceptible to humans. By applying the proposed principle, a neural network with human-understandable explanations is trained which, in a user study, is shown to better align with human intuition. This is likely to make the AI more trustworthy and opens the door to humans learning from machines. In the case study, we specifically investigate and analyze the behaviour of the human-imperceptible high spatial frequency features in neural networks and XAI methods.

Researchers and practitioners increasingly consider a human-centered perspective in the design of machine learning-based applications, especially in the context of Explainable Artificial Intelligence (XAI). However, clear methodological guidance in this context is still missing because each new situation seems to require a new setup, which also creates different methodological challenges. Existing case study collections in XAI inspired us; therefore, we propose a similar collection of case studies for human-centered XAI that can provide methodological guidance or inspiration for others. We want to showcase our idea in this workshop by describing three case studies from our research. These case studies are selected to highlight how apparently small differences require a different set of methods and considerations. With this workshop contribution, we would like to engage in a discussion on how such a collection of case studies can provide a methodological guidance and critical reflection.