Research Papers

Interpretable pipelines with evolutionarily optimized modules for RL tasks with visual inputs

Leonardo Lucio Custode, Giovanni Iacca

The importance of explainability in AI has become a pressing concern, for which several explainable AI (XAI) approaches have been recently proposed. However, most of the available XAI techniques are post-hoc methods, which however may be only partially reliable, as they do not reflect exactly the state of the original models. Thus, a more direct way for achieving XAI is through interpretable (also called glass-box) models. These models have been shown to obtain comparable (and, in some cases, better) performance with respect to black-boxes models in various tasks such as classification and reinforcement learning. However, they struggle when working with raw data, especially when the input dimensionality increases and the raw inputs alone do not give valuable insights on the decision-making process. Here, we propose to use end-to-end pipelines composed of multiple interpretable models co-optimized by means of evolutionary algorithms, that allows us to decompose the decision-making process into two parts: computing high-level features from raw data, and reasoning on the extracted high-level features. We test our approach in reinforcement learning environments from the Atari benchmark, where we obtain comparable results (with respect to black-box approaches) in settings without stochastic frame-skipping, while performance degrades in frame-skipping settings.

There has been recent and growing interest in the development and deployment of autonomous vehicles, encouraged by the empirical successes of powerful artificial intelligence techniques (AI), especially in the applications of deep learning and reinforcement learning. However, as demonstrated by recent traffic accidents, autonomous driving technology is not fully reliable for safe deployment. As AI is the main technology behind the intelligent navigation systems of self-driving vehicles, both the stakeholders and transportation regulators require their AI-driven software architecture to be safe, explainable, and regulatory compliant. In this paper, we propose a design framework that integrates autonomous control, explainable AI (XAI), and regulatory compliance to address this issue, and then provide an initial validation of the framework with a critical analysis in a case study. Moreover, we describe relevant XAI approaches that can help achieve the goals of the framework.

Vacuum Insulated Glazing (VIG) is a highly thermally insulating window technology, which boasts an extremely thin profile and lower weight as compared to gas-filled insulated glazing units of equivalent performance. The VIG is a double-pane configuration with a submillimeter vacuum gap between the panes and therefore under constant atmospheric pressure over their service life. Small pillars are positioned between the panes to maintain the gap, which can damage the glass reducing the lifetime of the VIG unit. To efficiently assess any surface damage on the glass, an automated damage detection system is highly desirable. For the purpose of classifying the damage, we have developed, trained, and tested a deep learning computer vision system using convolutional neural networks. The classification model flawlessly classified the test dataset with an area under the curve (AUC) for the receiver operating characteristic (ROC) of 100%. We have automatically cropped the images down to their relevant information by using Faster-RCNN to locate the position of the pillars. We employ the state-of-the-art methods Grad-CAM and Score-CAM of explainable Artificial Intelligence (XAI) to provide an understanding of the internal mechanisms and were able to show that our classifier outperforms ResNet50V2 for identification of crack locations and geometry. The proposed methods can therefore be used to detect systematic defects even without large amounts of training data. Further analyses of our model's predictive capabilities demonstrates its superiority over state-of-the-art models (ResNet50V2, ResNet101V2 and ResNet152V2) in terms of convergence speed, accuracy, precision at 100% recall and AUC for ROC.

While recent advances in AI-based automated decision-making have shown many benefits for businesses and society, they also come at a cost. It has for long been known that a high level of automation of decisions can lead to various drawbacks, such as automation bias and deskilling. In particular, the deskilling of knowledge workers is a major issue, as they are the same people who should also train, challenge and evolve AI. To address this issue, we conceptualize a new class of DSS, namely Intelligent Decision Assistance (IDA) based on a literature review of two different research streams -- DSS and automation. IDA supports knowledge workers without influencing them through automated decision-making. Specifically, we propose to use techniques of Explainable AI (XAI) while withholding concrete AI recommendations. To test this conceptualization, we develop hypotheses on the impacts of IDA and provide first evidence for their validity based on empirical studies in the literature.

A particular class of Explainable AI (XAI) methods provide saliency maps to highlight part of the image a Convolutional Neural Network (CNN) model looks at to classify the image as a way to explain its working. These methods provide an intuitive way for users to understand predictions made by CNNs. Other than quantitative computational tests, the vast majority of evidence to highlight that the methods are valuable is anecdotal. Given that humans would be the end-users of such methods, we devise three human subject experiments through which we gauge the effectiveness of these saliency-based explainability methods.

Explainable AI has attracted much research attention in recent years with feature attribution algorithms, which compute "feature importance" in predictions, becoming increasingly popular. However, there is little analysis of the validity of these algorithms as there is no "ground truth" in the existing datasets to validate their correctness. In this work, we develop a method to quantitatively evaluate the correctness of XAI algorithms by creating datasets with known explanation ground truth. To this end, we focus on the binary classification problems. String datasets are constructed using formal language derived from a grammar. A string is positive if and only if a certain property is fulfilled. Symbols serving as explanation ground truth in a positive string are part of an explanation if and only if they contributes to fulfilling the property. Two popular feature attribution explainers, Local Interpretable Model-agnostic Explanations (LIME) and SHapley Additive exPlanations (SHAP), are used in our experiments.We show that: (1) classification accuracy is positively correlated with explanation accuracy; (2) SHAP provides more accurate explanations than LIME; (3) explanation accuracy is negatively correlated with dataset complexity.

Enabling Machine Learning Algorithms for Credit Scoring -- Explainable Artificial Intelligence (XAI) methods for clear understanding complex predictive models

Przemysław Biecek, Marcin Chlebus, Janusz Gajda +5

Rapid development of advanced modelling techniques gives an opportunity to develop tools that are more and more accurate. However as usually, everything comes with a price and in this case, the price to pay is to loose interpretability of a model while gaining on its accuracy and precision. For managers to control and effectively manage credit risk and for regulators to be convinced with model quality the price to pay is too high. In this paper, we show how to take credit scoring analytics in to the next level, namely we present comparison of various predictive models (logistic regression, logistic regression with weight of evidence transformations and modern artificial intelligence algorithms) and show that advanced tree based models give best results in prediction of client default. What is even more important and valuable we also show how to boost advanced models using techniques which allow to interpret them and made them more accessible for credit risk practitioners, resolving the crucial obstacle in widespread deployment of more complex, 'black box' models like random forests, gradient boosted or extreme gradient boosted trees. All this will be shown on the large dataset obtained from the Polish Credit Bureau to which all the banks and most of the lending companies in the country do report the credit files. In this paper the data from lending companies were used. The paper then compares state of the art best practices in credit risk modelling with new advanced modern statistical tools boosted by the latest developments in the field of interpretability and explainability of artificial intelligence algorithms. We believe that this is a valuable contribution when it comes to presentation of different modelling tools but what is even more important it is showing which methods might be used to get insight and understanding of AI methods in credit risk context.

Algorithmic bias mitigation has been one of the most difficult conundrums for the data science community and Machine Learning (ML) experts. Over several years, there have appeared enormous efforts in the field of fairness in ML. Despite the progress toward identifying biases and designing fair algorithms, translating them into the industry remains a major challenge. In this paper, we present the initial results of an industrial open innovation project in the banking sector: we propose a general roadmap for fairness in ML and the implementation of a toolkit called BeFair that helps to identify and mitigate bias. Results show that training a model without explicit constraints may lead to bias exacerbation in the predictions.

According to a mainstream position in contemporary cognitive science and philosophy, the use of abstract compositional concepts is both a necessary and a sufficient condition for the presence of genuine thought. In this article, we show how the ability to develop and utilise abstract conceptual structures can be achieved by a particular kind of learning agents. More specifically, we provide and motivate a concrete operational definition of what it means for these agents to be in possession of abstract concepts, before presenting an explicit example of a minimal architecture that supports this capability. We then proceed to demonstrate how the existence of abstract conceptual structures can be operationally useful in the process of employing previously acquired knowledge in the face of new experiences, thereby vindicating the natural conjecture that the cognitive functions of abstraction and generalisation are closely related. Keywords: concept formation, projective simulation, reinforcement learning, transparent artificial intelligence, theory formation, explainable artificial intelligence (XAI)

Experiential AI

Drew Hemment, Ruth Aylett, Vaishak Belle +5

Experiential AI is proposed as a new research agenda in which artists and scientists come together to dispel the mystery of algorithms and make their mechanisms vividly apparent. It addresses the challenge of finding novel ways of opening up the field of artificial intelligence to greater transparency and collaboration between human and machine. The hypothesis is that art can mediate between computer code and human comprehension to overcome the limitations of explanations in and for AI systems. Artists can make the boundaries of systems visible and offer novel ways to make the reasoning of AI transparent and decipherable. Beyond this, artistic practice can explore new configurations of humans and algorithms, mapping the terrain of inter-agencies between people and machines. This helps to viscerally understand the complex causal chains in environments with AI components, including questions about what data to collect or who to collect it about, how the algorithms are chosen, commissioned and configured or how humans are conditioned by their participation in algorithmic processes.

Synthetic personae experiments have become a prominent method in Large Language Model alignment research, yet the representativeness and ecological validity of these personae vary considerably between studies. Through a review of 63 peer-reviewed studies published between 2023 and 2025 in leading NLP and AI venues, we reveal a critical gap: task and population of interest are often underspecified in persona-based experiments, despite personalization being fundamentally dependent on these criteria. Our analysis shows substantial differences in user representation, with most studies focusing on limited sociodemographic attributes and only 35% discussing the representativeness of their LLM personae. Based on our findings, we introduce a persona transparency checklist that emphasizes representative sampling, explicit grounding in empirical data, and enhanced ecological validity. Our work provides both a comprehensive assessment of current practices and practical guidelines to improve the rigor and ecological validity of persona-based evaluations in language model alignment research.

LLM-based agents have emerged as transformative tools capable of executing complex tasks through iterative planning and action, achieving significant advancements in understanding and addressing user needs. Yet, their effectiveness remains limited in specialized domains such as mental health diagnosis, where they underperform compared to general applications. Current approaches to integrating diagnostic capabilities into LLMs rely on scarce, highly sensitive mental health datasets, which are challenging to acquire. These methods also fail to emulate clinicians' proactive inquiry skills, lack multi-turn conversational comprehension, and struggle to align outputs with expert clinical reasoning. To address these gaps, we propose DSM5AgentFlow, the first LLM-based agent workflow designed to autonomously generate DSM-5 Level-1 diagnostic questionnaires. By simulating therapist-client dialogues with specific client profiles, the framework delivers transparent, step-by-step disorder predictions, producing explainable and trustworthy results. This workflow serves as a complementary tool for mental health diagnosis, ensuring adherence to ethical and legal standards. Through comprehensive experiments, we evaluate leading LLMs across three critical dimensions: conversational realism, diagnostic accuracy, and explainability. Our datasets and implementations are fully open-sourced.