Research Papers

On the Trade-offs between Adversarial Robustness and Actionable Explanations

Satyapriya Krishna, Chirag Agarwal, Himabindu Lakkaraju

As machine learning models are increasingly being employed in various high-stakes settings, it becomes important to ensure that predictions of these models are not only adversarially robust, but also readily explainable to relevant stakeholders. However, it is unclear if these two notions can be simultaneously achieved or if there exist trade-offs between them. In this work, we make one of the first attempts at studying the impact of adversarially robust models on actionable explanations which provide end users with a means for recourse. We theoretically and empirically analyze the cost (ease of implementation) and validity (probability of obtaining a positive model prediction) of recourses output by state-of-the-art algorithms when the underlying models are adversarially robust vs. non-robust. More specifically, we derive theoretical bounds on the differences between the cost and the validity of the recourses generated by state-of-the-art algorithms for adversarially robust vs. non-robust linear and non-linear models. Our empirical results with multiple real-world datasets validate our theoretical results and show the impact of varying degrees of model robustness on the cost and validity of the resulting recourses. Our analyses demonstrate that adversarially robust models significantly increase the cost and reduce the validity of the resulting recourses, thus shedding light on the inherent trade-offs between adversarial robustness and actionable explanations.

The significant features identified in a representative subset of the dataset during the learning process of an artificial intelligence model are referred to as a 'global' explanation. 3D global explanations are crucial in neuroimaging, where a complex representational space demands more than basic 2D interpretations. However, current studies in the literature often lack the accuracy, comprehensibility, and 3D global explanations needed in neuroimaging and beyond. To address this gap, we developed an explainable artificial intelligence (XAI) 3D-Framework capable of providing accurate, low-complexity global explanations. We evaluated the framework using various 3D deep learning models trained on a well-annotated cohort of 596 structural MRIs. The binary classification task focused on detecting the presence or absence of the paracingulate sulcus, a highly variable brain structure associated with psychosis. Our framework integrates statistical features (Shape) and XAI methods (GradCam and SHAP) with dimensionality reduction, ensuring that explanations reflect both model learning and cohort-specific variability. By combining Shape, GradCam, and SHAP, our framework reduces inter-method variability, enhancing the faithfulness and reliability of global explanations. These robust explanations facilitated the identification of critical sub-regions, including the posterior temporal and internal parietal regions, as well as the cingulate region and thalamus, suggesting potential genetic or developmental influences. Our XAI 3D-Framework leverages global explanations to uncover the broader developmental context of specific cortical features. This approach advances the fields of deep learning and neuroscience by offering insights into normative brain development and atypical trajectories linked to mental illness, paving the way for more reliable and interpretable AI applications in neuroimaging.

Foundation models including large language models (LLMs) are increasingly attracting interest worldwide for their distinguished capabilities and potential to perform a wide variety of tasks. Nevertheless, people are concerned about whether foundation model based AI systems are properly governed to ensure the trustworthiness and to prevent misuse that could harm humans, society and the environment. In this paper, we identify eight governance challenges of foundation model based AI systems regarding the three fundamental dimensions of governance: decision rights, incentives, and accountability. Furthermore, we explore the potential of blockchain as an architectural solution to address the challenges by providing a distributed ledger to facilitate decentralised governance. We present an architecture that demonstrates how blockchain can be leveraged to realise governance in foundation model based AI systems.

To harness the true potential of Artificial Intelligence (AI) for societal betterment, we need to move away from prioritising corporate interests which exploit Global South workers in the digital age. The unpaid labour and societal harms which are generated by Digital Value Networks (DVNs) disproportionately affect workers in Africa, Latin America, and India and need to be regulated. In this research, we discuss unethical practices to automate Human Intelligence Tasks (HITs) through gig work platforms and the capitalisation of data collection utilising influencers in social media. These are important areas of study in worker and user data practices, where ethical AI could be impactful. We provide suggestions for a path forward focused on responsible AI development.

Neural network models are widely used in a variety of domains, often as black-box solutions, since they are not directly interpretable for humans. The field of explainable artificial intelligence aims at developing explanation methods to address this challenge, and several approaches have been developed over the recent years, including methods for investigating what type of knowledge these models internalise during the training process. Among these, the method of concept detection, investigates which \emph{concepts} neural network models learn to represent in order to complete their tasks. In this work, we present an extension to the method of concept detection, named \emph{concept backpropagation}, which provides a way of analysing how the information representing a given concept is internalised in a given neural network model. In this approach, the model input is perturbed in a manner guided by a trained concept probe for the described model, such that the concept of interest is maximised. This allows for the visualisation of the detected concept directly in the input space of the model, which in turn makes it possible to see what information the model depends on for representing the described concept. We present results for this method applied to a various set of input modalities, and discuss how our proposed method can be used to visualise what information trained concept probes use, and the degree as to which the representation of the probed concept is entangled within the neural network model itself.

Mitigating Label Bias via Decoupled Confident Learning

Yunyi Li, Maria De-Arteaga, Maytal Saar-Tsechansky

Growing concerns regarding algorithmic fairness have led to a surge in methodologies to mitigate algorithmic bias. However, such methodologies largely assume that observed labels in training data are correct. This is problematic because bias in labels is pervasive across important domains, including healthcare, hiring, and content moderation. In particular, human-generated labels are prone to encoding societal biases. While the presence of labeling bias has been discussed conceptually, there is a lack of methodologies to address this problem. We propose a pruning method -- Decoupled Confident Learning (DeCoLe) -- specifically designed to mitigate label bias. After illustrating its performance on a synthetic dataset, we apply DeCoLe in the context of hate speech detection, where label bias has been recognized as an important challenge, and show that it successfully identifies biased labels and outperforms competing approaches.

Our work serves as a framework for unifying the challenges of contemporary explainable AI (XAI). We demonstrate that while XAI methods provide supplementary and potentially useful output for machine learning models, researchers and decision-makers should be mindful of their conceptual and technical limitations, which frequently result in these methods themselves becoming black boxes. We examine three XAI research avenues spanning image, textual, and graph data, covering saliency, attention, and graph-type explainers. Despite the varying contexts and timeframes of the mentioned cases, the same persistent roadblocks emerge, highlighting the need for a conceptual breakthrough in the field to address the challenge of compatibility between XAI methods and application tasks.

A novel structured argumentation framework for improved explainability of classification tasks

Lucas Rizzo, Luca Longo

This paper presents a novel framework for structured argumentation, named extend argumentative decision graph ($xADG$). It is an extension of argumentative decision graphs built upon Dung's abstract argumentation graphs. The $xADG$ framework allows for arguments to use boolean logic operators and multiple premises (supports) within their internal structure, resulting in more concise argumentation graphs that may be easier for users to understand. The study presents a methodology for construction of $xADGs$ and evaluates their size and predictive capacity for classification tasks of varying magnitudes. Resulting $xADGs$ achieved strong (balanced) accuracy, which was accomplished through an input decision tree, while also reducing the average number of supports needed to reach a conclusion. The results further indicated that it is possible to construct plausibly understandable $xADGs$ that outperform other techniques for building $ADGs$ in terms of predictive capacity and overall size. In summary, the study suggests that $xADG$ represents a promising framework to developing more concise argumentative models that can be used for classification tasks and knowledge discovery, acquisition, and refinement.

In the domain of unsupervised image-to-image transformation using generative transformative models, CycleGAN has become the architecture of choice. One of the primary downsides of this architecture is its relatively slow rate of convergence. In this work, we use discriminator-driven explainability to speed up the convergence rate of the generative model by using saliency maps from the discriminator that mask the gradients of the generator during backpropagation, based on the work of Nagisetty et al., and also introducing the saliency map on input, added onto a Gaussian noise mask, by using an interpretable latent variable based on Wang M.'s Mask CycleGAN. This allows for an explainability fusion in both directions, and utilizing the noise-added saliency map on input as evidence-based counterfactual filtering. This new architecture has much higher rate of convergence than a baseline CycleGAN architecture while preserving the image quality.

With the growing concern for air quality and its impact on human health, interest in environmental gas monitoring has increased. However, chemi-resistive gas sensing devices are plagued by issues of sensor reproducibility during manufacturing. This study proposes a novel approach for detecting sensor-to-sensor variations in sensing devices using the explainable AI (XAI) method of SHapley Additive exPlanations (SHAP). This is achieved by identifying sensors that contribute the most to environmental gas concentration estimation via machine learning, and measuring the similarity of feature rankings between sensors to flag deviations or outliers. The methodology is tested using artificial and realistic Ozone concentration profiles to train a Gated Recurrent Unit (GRU) model. Two applications were explored in the study: the detection of wrong behaviors of sensors in the train dataset, and the detection of deviations in the test dataset. By training the GRU with the pruned train dataset, we could reduce computational costs while improving the model performance. Overall, the results show that our approach improves the understanding of sensor behavior, successfully detects sensor deviations down to 5-10% from the normal behavior, and leads to more efficient model preparation and calibration. Our method provides a novel solution for identifying deviating sensors, linking inconsistencies in hardware to sensor-to-sensor variations in the manufacturing process on an AI model-level.

Progress in graph neural networks has grown rapidly in recent years, with many new developments in drug discovery, medical diagnosis, and recommender systems. While this progress is significant, many networks are `black boxes' with little understanding of the `what' exactly the network is learning. Many high-stakes applications, such as drug discovery, require human-intelligible explanations from the models so that users can recognize errors and discover new knowledge. Therefore, the development of explainable AI algorithms is essential for us to reap the benefits of AI. We propose an explainability algorithm for GNNs called eXplainable Insight (XInsight) that generates a distribution of model explanations using GFlowNets. Since GFlowNets generate objects with probabilities proportional to a reward, XInsight can generate a diverse set of explanations, compared to previous methods that only learn the maximum reward sample. We demonstrate XInsight by generating explanations for GNNs trained on two graph classification tasks: classifying mutagenic compounds with the MUTAG dataset and classifying acyclic graphs with a synthetic dataset that we have open-sourced. We show the utility of XInsight's explanations by analyzing the generated compounds using QSAR modeling, and we find that XInsight generates compounds that cluster by lipophilicity, a known correlate of mutagenicity. Our results show that XInsight generates a distribution of explanations that uncovers the underlying relationships demonstrated by the model. They also highlight the importance of generating a diverse set of explanations, as it enables us to discover hidden relationships in the model and provides valuable guidance for further analysis.

In this survey paper, we deep dive into the field of Explainable Artificial Intelligence (XAI). After introducing the scope of this paper, we start by discussing what an "explanation" really is. We then move on to discuss some of the existing approaches to XAI and build a taxonomy of the most popular methods. Next, we also look at a few applications of these and other XAI techniques in four primary domains: finance, autonomous driving, healthcare and manufacturing. We end by introducing a promising discipline, "Explanation Engineering," which includes a systematic approach for designing explainability into AI systems.