State-of-the-art imitation learning from observation methods (ILfO) have recently made significant progress, but they still have some limitations: they need action-based supervised optimisation, assume that states have a single optimal action, and tend to apply teacher actions without full consideration of the actual environment state. While the truth may be out there in observed trajectories, existing methods struggle to extract it without supervision. In this work, we propose Unsupervised Imitation Learning from Observation (UfO) that addresses all of these limitations. UfO learns a policy through a two-stage process, in which the agent first obtains an approximation of the teacher's true actions in the observed state transitions, and then refines the learned policy further by adjusting agent trajectories to closely align them with the teacher's. Experiments we conducted in five widely used environments show that UfO not only outperforms the teacher and all other ILfO methods but also displays the smallest standard deviation. This reduction in standard deviation indicates better generalisation in unseen scenarios.

Online goal recognition in continuous domains poses two central challenges: efficiently encoding large trajectories and effectively comparing them. Recent work addresses these challenges by using custom state-space representations and metrics to compare observations against hypotheses. However, these approaches often overlook well-established encoding techniques used in other domains that offer substantial advantages. This paper introduces a novel method for online goal recognition that leverages path signatures, a compact, expressive representation of rough path theory that captures key semantic features of trajectories in an efficient way, enabling more meaningful comparisons between them. Experiments show that our method consistently outperforms the state of the art in predictive accuracy and online planning efficiency, while remaining competitive offline.

Reinforcement Learning (RL) policies often degrade in unfamiliar environments because they lack explicit deliberation. We propose Plan, Align, Commit, Think (PACT), a hybrid architecture that combines a fast, reactive RL policy with a slow, deliberative Small Language Model (SLM) planner. PACT invokes the SLM asynchronously to generate and validate candidate action plans. Once a plan is verified through simulation as safe, feasible, and complete, it is executed directly, bypassing the RL policy without retraining or modifying it. Evaluated on three FrozenLake configurations of increasing difficulty, PACT outperforms all baselines while relying on a 2B-parameter SLM backbone, suggesting that deliberative planning and reactive execution are more powerful in concert than either is alone in these settings.

Wikidata is one of the world's most important data assets. It is used by search engines, virtual assistants, fact checkers, and in over 800 Wikimedia projects. Wikidata contains 1.65 billion statements about over 116 million data items, edited by nearly 25 thousand editors. Manually checking whether Wikidata statements are supported by references is a slow process that does not scale with size. Given the overall number of statements to check, the collaborative nature of Wikidata, and the fact that referenced documents can change over time, preserving the quality of the references is an onerous process requiring continuous intervention. This paper present ProVe – a tool to assist in the automatic verification and assessment of the quality of the references of Wikidata items.

Reinforcement learning agents operating under partial observability must act on incomplete information, making them natural candidates for guidance from small language models (SLMs) that carry broad reasoning priors. Yet integrating SLM guidance into this setting has proven difficult: across all test environments, vanilla uncertainty-gated approaches achieve an overwrite rate at or near zero, meaning the SLM almost never contributes an independent action. We trace this failure to the bare egocentric prompt, which provides insufficient context for genuine reasoning, and identify it as a context problem rather than a capacity problem. We propose ASK+, which supplies the SLM with trajectory-aware context (a partially revealed map, visited positions, and action history) and structured chain-of-thought reasoning, converting it from a passive redundancy check into a more informative consultant that occasionally corrects the policy. We further establish that the predictive entropy signal used for selective querying measures action uncertainty rather than state uncertainty and remains informative in POMDPs, making uncertainty-gated assistance viable beyond fully observable settings. The stateful prompt drives substantial gains: on DoorKey, where vanilla ASK matches PPO (both 89

Imitation learning stands at a crossroads: despite decades of progress, current imitation learning agents remain sophisticated memorisation machines, excelling at replay but failing when contexts shift or goals evolve. This paper argues that this failure is not technical but foundational: imitation learning has been optimised for the wrong objective. We propose a research agenda that redefines success from perfect replay to compositional adaptability. Such adaptability hinges on learning behavioural primitives once and recombining them through novel contexts without retraining. We establish metrics for compositional generalisation, propose hybrid architectures, and outline interdisciplinary research directions drawing on cognitive science and cultural evolution. Agents that embed adaptability at the core of imitation learning thus have an essential capability for operating in an open-ended world.

The topic of forgetting, which loosely speaking means losing, removing, or even hiding some variables, propositions, or formulas, has been extensively studied in the field of knowledge representation and reasoning for many major formalisms. In this article, we convey this topic to the highly active field of abstract argumentation. We provide an in-depth analysis of desirable syntactical and/or semantical properties of possible forgetting operators. In doing so, we included well-known logic programming conditions, such as strong persistence or strong invariance. Further, we argue that although abstract argumentation and logic programming are closely related, it is not possible to reduce forgetting in abstract argumentation to forgetting in logic programming in a straightforward manner. The analysis of desiderata, adapted to the specifics of abstract argumentation, includes implications among them, individual and collective satisfiability, and identifying inherent limits for a set of prominent semantics. Finally, we conduct a case study on stable semantics incorporating concrete forgetting operators.

We present ICX (Individual Consistency eXplorer), an interactive dashboard designed to support stakeholders in exploring individual fairness notions within algorithmic decision-making systems. ICX focuses on a set of metrics based on the consistency score, a key measure of individual fairness, by allowing the visualisation of how the classification of an individual compares with that of similar individuals. Stakeholders can define and fine-tune the notion of similarity according to domain-specific criteria, and examine individual-level views that highlight comparable individuals and their classification outcomes. ICX empowers users to interrogate, analyse and interpret fairness at the individual level, making algorithmic decision-making more transparent and accountable.

Imitation learning benchmarks often lack sufficient variation between training and evaluation, limiting meaningful generalisation assessment. We introduce Labyrinth, a benchmarking environment designed to test generalisation with precise control over structure, start and goal positions, and task complexity. It enables verifiably distinct training, evaluation, and test settings. Labyrinth provides a discrete, fully observable state space and known optimal actions, supporting interpretability and fine-grained evaluation. Its flexible setup allows targeted testing of generalisation factors and includes variants like partial observability, key-and-door tasks, and ice-floor hazards. By enabling controlled, reproducible experiments, Labyrinth advances the evaluation of generalisation in imitation learning and provides a valuable tool for developing more robust agents.

Individual fairness is the principle aiming for equitable treatment for each individual affected by decisions. Despite its intuitive appeal, the practical applications of individual fairness for algorithmic decision-making systems remain relatively unexplored. In this paper, we investigate the consistency score metric and demonstrate how it fails to adequately capture fairness at the individual level, underscoring the need for a more fine-grained approach. We show that (1) the consistency score obscures instances where individuals are treated significantly differently to the individuals most similar to them and (2) the perceived fairness of individual decisions can be affected by several factors, including the similarity notion itself. To address these issues, we propose four new metrics that measure different aspects of the treatment of individuals with respect to similar individuals, under varying similarity definitions. Our comprehensive evaluation of the new metrics shows that they offer a more nuanced approach to assessing individual fairness, enabling decision-makers to focus on individuals most adversely affected by controversial decisions.

As algorithmic decision-making systems become more prevalent in society, ensuring the fairness of these systems is becoming increasingly important. Whilst there has been substantial research in building fair algorithmic decision-making systems, the majority of these methods require access to the training data, including personal characteristics, and are not transparent regarding which individuals are classified unfairly. In this paper, we propose a novel model-agnostic argumentation-based method to determine why an individual is classified differently in comparison to similar individuals. Our method uses a quantitative argumentation framework to represent attribute-value pairs of an individual and of those similar to them, and uses a well-known semantics to identify the attribute-value pairs in the individual contributing most to their different classification. We evaluate our method on two datasets commonly used in the fairness literature and illustrate its effectiveness in the identification of bias.

Some imitation learning methods combine behavioural cloning with self-supervision to infer actions from state pairs. However, most rely on a large number of expert trajectories to increase generalisation and human intervention to capture key aspects of the problem, such as domain constraints. In this paper, we propose Continuous Imitation Learning from Observation (CILO), a new method augmenting imitation learning with two important features: (i) exploration, allowing for more diverse state transitions, requiring less expert trajectories and resulting in fewer training iterations; and (ii) path signatures, allowing for automatic encoding of constraints, through the creation of non-parametric representations of agents and expert trajectories. We compared CILO with a baseline and two leading imitation learning methods in five environments. It had the best overall performance of all methods in all environments, outperforming the expert in two of them.

As algorithmic decision-making systems (ADMS) are increasingly deployed across various sectors, the importance of research on fairness in Artificial Intelligence (AI) continues to grow. In this paper we highlight a number of significant practical limitations and regulatory compliance issues associated with the application of existing bias mitigation methods to ADMS. We present an example of an algorithmic system used in recruitment to illustrate these limitations. Our analysis of existing methods indicates a pressing need for a change in the approach to the development of new methods. In order to address the limitations, we provide recommendations for key factors to consider in the development of new bias mitigation methods that aim to be effective in real-world scenarios and comply with non-discrimination, data protection and sector-specific regulations. Further, we suggest a checklist relating to these recommendations that should be included with the development of new bias mitigation methods.

Theory of Mind capabilities, i.e. the capacity to adopt and reason from the perspective of others. By combining the Theory of Mind of TomAbd agents with abductive reasoning, agents can infer explanations for the behaviour of others, which they can incorporate into their own decision-making. We have implemented the TomAbd agent model and successfully tested its performance in the cooperative board game Hanabi.

Imitation learning is an approach in which an agent learns how to execute a task by trying to mimic how one or more teachers perform it. This learning approach offers a compromise between the time it takes to learn a new task and the effort needed to collect teacher samples for the agent. It achieves this by balancing learning from the teacher, who has some information on how to perform the task, and deviating from their examples when necessary, such as states not present in the teacher samples. Consequently, the field of imitation learning has received much attention from researchers in recent years, resulting in many new methods and applications. However, with this increase in published work and past surveys focusing mainly on methodology, a lack of standardisation became more prominent in the field. This non-standardisation is evident in the use of environments, which appear in no more than two works, and evaluation processes, such as qualitative analysis, that have become rare in current literature. In this survey, we systematically review current imitation learning literature and present our findings by (i) classifying imitation learning techniques, environments and metrics by introducing novel taxonomies; (ii) reflecting on main problems from the literature; and (iii) presenting challenges and future directions for researchers.

Knowledge Graphs are repositories of information that gather data from a multitude of domains and sources in the form of semantic triples, serving as a source of structured data for various crucial applications in the modern web landscape, from Wikipedia infoboxes to search engines. Such graphs mainly serve as secondary sources of information and depend on well-documented and verifiable provenance to ensure their trustworthiness and usability. However, their ability to systematically assess and assure the quality of this provenance, most crucially whether it properly supports the graph’s information, relies mainly on manual processes that do not scale with size. ProVe aims at remedying this, consisting of a pipelined approach that automatically verifies whether a Knowledge Graph triple is supported by text extracted from its documented provenance. ProVe is intended to assist information curators and consists of four main steps involving rule-based methods and machine learning models: text extraction, triple verbalisation, sentence selection, and claim verification. ProVe is evaluated on a Wikidata dataset, achieving promising results overall and excellent performance on the binary classification task of detecting support from provenance, with 87.5 % accuracy and 82.9 % F1-macro on text-rich sources. The evaluation data and scripts used in this paper are available in GitHub and Figshare.

Imitation learning field requires expert data to train agents in a task. Most often, this learning approach suffers from the absence of available data, which results in techniques being tested on its dataset. Creating datasets is a cumbersome process requiring researchers to train expert agents from scratch, record their interactions and test each benchmark method with newly created data. Moreover, creating new datasets for each new technique results in a lack of consistency in the evaluation process since each dataset can drastically vary in state and action distribution. In response, this work aims to address these issues by creating Imitation Learning Datasets, a toolkit that allows for: (i) curated expert policies with multithreaded support for faster dataset creation; (ii) readily available datasets and techniques with precise measurements; and (iii) sharing implementations of common imitation learning techniques. Demonstration link: https://nathangavenski.github.io/#/il-datasets-video

With AI-based decision-making systems increasingly being deployed in various sectors, research on fairness in AI has become even more important. In this position paper, we highlight a number of significant practical applicability limitations and regulatory compliance issues associated with existing bias mitigation methods. These limitations indicate a pressing need for a change in the approach to their development. In order to address them, we provide a list of recommendations for new bias mitigation methods that are not only effective, but can also be applied in real-world scenarios and comply with legal requirements.

This paper proposes a canonical ordering of arguments within abstract argumentation labellings and two new types of efficient representations of these labellings for use in applications involving the computation of argumentation semantics. The space requirements of the representations are analysed, benchmarked on a class of hard enumeration problems taken from the International Competition on Computational Models of Argumentation (ICCMA), and compared for efficiency. We found that they both offer significant reductions of the memory representation requirements of large labellings, sometimes of up to 75%. We argue that the new way of looking at labellings provided by one of the representations, i.e., by considering repetitions of segment assignments within labellings, paves the way for investigations of new applications in argumentation theory.

Algorithmic Decision-Making Systems (ADMS) 1 fairness issues have been well highlighted over the past decade [1] , including some facial recognition systems struggling to identify people of color [2] . In 2021, Uber drivers filed a claim with the U.K. ’s employment tribunal for unfair dismissal resulting from automated facial recognition technology by Microsoft [3] . Bias mitigation methods have been developed to reduce discrimination from ADMS. These typically operationalize fairness notions as fairness metrics to minimize discrimination [4] . We refer to ADMS to which bias mitigation methods have been applied as “mitigated ADMS” or, in the singular, a “mitigated system.”