Results
- Showing results for:
- Reset all filters
Search results
-
Conference paperJiang J, Bewley T, Amoukou S, et al., 2025,
Representation consistency for accurate and coherent LLM answer aggregation
, The Thirty-Ninth Annual Conference on Neural Information Processing Systems, Publisher: Neural Information Processing Systems Foundation, Inc. (NeurIPS)Test-time scaling improves large language models' (LLMs) performance by allocating more compute budget during inference. To achieve this, existing methods often require intricate modifications to prompting and sampling strategies. In this work, we introduce representation consistency (RC), a test-time scaling method for aggregating answers drawn from multiple candidate responses of an LLM regardless of how they were generated, including variations in prompt phrasing and sampling strategy. RC enhances answer aggregation by not only considering the number of occurrences of each answer in the candidate response set, but also the consistency of the model's internal activations while generating the set of responses leading to each answer. These activations can be either dense (raw model activations) or sparse (encoded via pretrained sparse autoencoders). Our rationale is that if the model's representations of multiple responses converging on the same answer are highly variable, this answer is more likely to be the result of incoherent reasoning and should be down-weighted during aggregation. Importantly, our method only uses cached activations and lightweight similarity computations and requires no additional model queries. Through experiments with four open-source LLMs and four reasoning datasets, we validate the effectiveness of RC for improving task performance during inference, with consistent accuracy improvements (up to 4%) over strong test-time scaling baselines. We also show that consistency in the sparse activation signals aligns well with the common notion of coherent reasoning.
-
Conference paperGould A, Toni F, 2025,
Neuro-argumentative learning with case-based reasoning
, 19th International Conference on Neurosymbolic Learning and Reasoning, Publisher: MLResearchPress, Pages: 1090-1106We introduce Gradual Abstract Argumentation for Case-Based Reasoning (Gradual AA-CBR), a data-driven, neurosymbolic classification model in which the outcome is determined by an argumentation debate structure that is learned simultaneously with neural-based feature extractors. Each argument in the debate is an observed case from the training data, favouring their labelling. Cases attack or support those with opposing or agreeing labellings, with the strength of each argument and relationship learned through gradient-based methods. This argumentation debate structure provides human-aligned reasoning, improving model interpretability compared to traditional neural networks (NNs). Unlike the existing purely symbolic variant, Abstract Argumentation for Case-Based Reasoning (AA-CBR), Gradual AA-CBR is capable of multi-class classification, automatic learning of feature and data point importance, assigning uncertainty values to outcomes, using all available data points, and does not require binary features. We show that Gradual AA-CBR performs comparably to NNs whilst significantly outperforming existing AA-CBR formulations.
-
Conference paperJacob AR, Kori A, Angelis ED, et al., 2025,
Object-centric neuro-argumentative learning
, Conference on Neurosymbolic Learning and Reasoning, Publisher: MLResearchPress, Pages: 1077-1089Over the last decade, as we rely more on deep learning technologies to make critical decisions, concerns regarding their safety, reliability and interpretability have emerged. We introduce a novel Neural Argumentative Learning (NAL) architecture that integrates Assumption-Based Argumentation (ABA) with Object-Centric (OC) deep learning for im- age analysis. Our OC-NAL architecture consists of neural and symbolic components. The former segments and encodes images into facts, while the latter applies ABA learning to develop ABA frameworks enabling image classification. Experiments on synthetic data show that the OC-NAL architecture can be competitive with a state-of-the-art alternative. The code can be found at https://github.com/AbdulRJacob/Neuro-AL.
-
Conference paperBlümel L, Rapberger A, Thimm M, et al., 2025,
On independence and SCC-recursiveness in assumption-based argumentation
, Thirty-Fourth International Joint Conference on Artificial Intelligence {IJCAI-25}, Publisher: International Joint Conferences on Artificial Intelligence Organization, Pages: 4382-4390We introduce a notion of conditional independence in (flat) assumption-based argumentation (ABA), where independence between (sets of) assumptions amounts to the presence of information about one set of assumptions not impacting the acceptability of another. We study general properties, computational complexity, and the relation to independence in abstract argumentation. In light of the high computational complexity of deciding independence, we introduce sound methods for checking independence in polynomial time via two different routes: the first utilizes the strongly connected components (SCCs) of the instantiated abstract argumentation framework; the second exploits the structure of the ABA framework directly. Along the way, we introduce the notion of SCC-recursiveness for ABA.
-
Journal articleLeofante F, Artelt A, Eliades D, et al., 2025,
Explainable AI, energy and critical infrastructure systems
, AI Magazine, Vol: 46, ISSN: 0738-4602The AAAI 2025 Bridge on “Explainable AI, Energy and Critical Infrastructure Systems” was held at the Pennsylvania Convention Centre, Philadelphia, Pennsylvania, USA, on February 25, 2025. The bridge gathered researchers and practitioners, bringing together innovation research across explainable AI, energy and critical infrastructure systems so they can enhance each other. The Bridge featured five keynote presentations by experts, one tutorial, poster presentations by authors who contributed their research findings, and three breakout sessions to discuss new challenges arising at the intersection of these exciting disciplines.
-
Conference paperAlfano G, Gould A, Leofante F, et al., 2025,
Counterfactual explanations under model multiplicity and their use in computational argumentation
, International Joint Conference on Artificial Intelligence (IJCAI) 2025, Publisher: IJCAI, Pages: 4321-4329Counterfactual explanations (CXs) are widely recognised as an essential technique for providing recourse recommendations for AI models. However, it is not obvious how to determine CXs in model multiplicity scenarios, where equally performing but different models can be obtained for the same task. In this paper, we propose novel qualitative and quantitative definitions of CXs based on explicit, nested quantification over (groups) of model decisions. We also study properties of these notions and identify decision problems of interest therefor. While our CXs are broadly applicable, in this paper we instantiate them within computational argumentation where model multiplicity naturally emerges, e.g. with incomplete and case-based argumentation frameworks. We then illustrate the suitability of our CXs for model multiplicity in legal and healthcare contexts, before analysing the complexity of the associated decision problems.
-
Conference paperKori A, Toni F, Glocker B, 2025,
Identifiable object representations under spatial ambiguities
, International Conference on Machine Learning 2025, Publisher: MLResearchPress, Pages: 31486-31518, ISSN: 2640-3498Modular object-centric representations are essential for human-like reasoning but are challenging to obtain under spatial ambiguities, e.g. due to occlusions and view ambiguities. However, addressing challenges presents both theoretical and practical difficulties. We introduce a novel multi-view probabilistic approach that aggregates view-specific slots to capture invariant content information while simultaneously learning disentangled global viewpoint-level information. Unlike prior single-view methods, our approach resolves spatial ambiguities, provides theoretical guarantees for identifiability, and requires no viewpoint annotations. Extensive experiments on standard benchmarks and novel complex datasets validate our method’s robustness and scalability.
-
Conference paperRapberger A, Russo F, Rago A, et al., 2025,
On gradual semantics for assumption-based argumentation
, 22nd International Conference on Principles of Knowledge Representation and Reasoning (KR 2025)In computational argumentation, gradual semantics are fine-grained alternatives to extension-based and labelling-basedsemantics. They ascribe a dialectical strength to (components of) arguments sanctioning their degree of acceptability. Several gradual semantics have been studied for abstract, bipolar and quantitative bipolar argumentation frameworks (QBAFs), as well as, to a lesser extent, for some forms of structured argumentation. However, this has not been the case for assumption-based argumentation (ABA), despite it being a popular form of structured argumentation with several applications where gradual semantics could be useful. In this paper, we fill this gap and propose a family of novel gradual semantics for equipping assumptions, which are the core components in ABA frameworks, with dialectical strengths. To do so, we use bipolar set-based argumentation frame-works as an abstraction of (potentially non-flat) ABA frame-works and generalise state-of-the-art modular gradual semantics for QBAFs. We show that our gradual ABA semantics satisfy suitable adaptations of desirable properties of gradual QBAF semantics, such as balance and monotonicity. We also explore an argument-based approach that leverages established QBAF modular semantics directly, and use it as base-line. Finally, we conduct experiments with synthetic ABA frameworks to compare our gradual ABA semantics with its argument-based counterpart and assess convergence.
-
Journal articleDickie C, Lauren S, Belardinelli F, et al., 2025,
Aggregating bipolar opinions through bipolar assumption-based argumentation
, Autonomous Agents and Multi-Agent Systems, Vol: 39, ISSN: 1387-2532We introduce a novel method to aggregate Bipolar ArgumentationFrameworks expressing opinions of different parties in debates. We use BipolarAssumption-based Argumentation (ABA) as an all-encompassing formalismfor Bipolar Argumentation under different semantics. By leveraging on recentresults on judgement aggregation in Social Choice Theory, we prove severalpreservation results for relevant properties of Bipolar ABA using quota andoligarchic rules. Specifically, we prove (positive and negative) results about thepreservation of conflict-free, closed, admissible, preferred, complete, set-stable,well-founded and ideal extensions in Bipolar ABA, as well as the preservationof acceptability, acyclicity and coherence for individual assumptions. Finally,we illustrate our methodology and results in the context of a case study onopinion aggregation for the treatment of long COVID patients.
-
Conference paperFreedman G, Toni F, 2025,
Exploring the potential for large language models to demonstrate rational probabilistic beliefs
, 38th International FLAIRS Conference, Publisher: LibraryPress@UF, ISSN: 2334-0762Advances in the general capabilities of large language models (LLMs) have led to their use for information retrieval, and as components in automated decision systems. A faithful representation of probabilistic reasoning in these models may be essential to ensure trustworthy, explainable and effective performance in these tasks. Despite previous work suggesting that LLMs can perform complex reasoning and well-calibrated uncertainty quantification, we find that current versions of this class of model lack the ability to provide rational and coherent representations of probabilistic beliefs. To demonstrate this, we introduce a novel dataset of claims with indeterminate truth values and apply a number of well-established techniques for uncertainty quantification to measure the ability of LLM's to adhere to fundamental properties of probabilistic reasoning.
-
Conference paperChen L, Dejl A, Toni F, 2025,
Identifying Query-Relevant Neurons in Large Language Models for Long-Form Texts
, The 39th Annual AAAI Conference on Artificial Intelligence -
Conference paperFreedman G, Dejl A, Gorur D, et al., 2025,
Argumentative large language models for explainable and contestable claim verification
, AAAI Conference on Artificial Intelligence, Publisher: Association for the Advancement of Artificial Intelligence, Pages: 14930-14939, ISSN: 2159-5399The profusion of knowledge encoded in large language models (LLMs) and their ability to apply this knowledge zero-shot in a range of settings makes them promising candidates for use in decision-making. However, they are currently limited by their inability to provide outputs which can be faithfully explained and effectively contested to correct mistakes. In this paper, we attempt to reconcile these strengths and weaknesses by introducing argumentative LLMs (ArgLLMs), a method for augmenting LLMs with argumentative reasoning. Concretely, ArgLLMs construct argumentation frameworks, which then serve as the basis for formal reasoning in support of decision-making. The interpretable nature of these argumentation frameworks and formal reasoning means that any decision made by ArgLLMs may be explained and contested. We evaluate ArgLLMs’ performance experimentally in comparison with state-of-the-art techniques, in the context of the decision-making task of claim verification. We also define novel properties to characterise contestability and assess ArgLLMs formally in terms of these properties.
-
Conference paperAyoobi H, Potyka N, Toni F, 2025,
ProtoArgNet: Interpretable Image Classification with Super-Prototypes and Argumentation
, AAAI Conference on Artificial Intelligence -
Conference paperRusso F, Toni F, 2025,
Shapley-PC: constraint-based causal structure learning with a Shapley inspired framework
, 4th Conference on Causal Learning and Reasoning (CLeaR 2025)Causal Structure Learning (CSL), also referred to as causal discovery, amounts to extracting causal relations among variables in data. CSL enables the estimation of causal effects from observational data alone, avoiding the need to perform real life experiments. Constraint-based CSL leverages conditional independence tests to perform causal discovery. We propose Shapley-PC, a novel method to improve constraint-based CSL algorithms by using Shapley values over the possible conditioning sets, to decide which variables are responsible for the observed conditional (in)dependences. We prove soundness, completeness and asymptotic consistency of Shapley-PC and run a simulationstudy showing that our proposed algorithm is superior to existing versions of PC.
-
Conference paperKori A, Glocker B, Toni F, 2025,
Explaining Image Classifiers with Visual Debates
, 27th International Conference on Discovery Science, Publisher: SPRINGER INTERNATIONAL PUBLISHING AG, Pages: 200-214, ISSN: 2945-9133- Cite
- Citations: 1
-
Conference paperDe Angelis E, Proietti M, Toni F, 2025,
Greedy ABA Learning for Case-Based Reasoning
, 24th International Conference on Autonomous Agents and Multiagent Systems-AAMAS-Annual, Publisher: ASSOC COMPUTING MACHINERY, Pages: 556-564 -
Conference paperKori A, Rago A, Toni F, 2025,
Free argumentative exchanges for explaining image classifiers
, AAMAS 2025, Publisher: ACMDeep learning models are powerful image classifiers but their opacity hinders their trustworthiness. Explanation methods for capturing the reasoning process within these classifiers faithfully and in a cognitively manageable manner are scarce, due to their sheer complexity and size. In this paper, we provide a solution for this problem by defining a novel method for explaining the outputs of image classifiers with debates between two agents, each arguing for a particular class. We obtain these debates as concrete instances of Free Argumentative eXchanges (FAXs), a novel argumentation-based multi-agent framework allowing agents to internalise opinions by other agents differently than originally stated. We define two metrics to assess the usefulness of FAXs as argumentative explanationsfor image classifiers. We then conduct a number of empirical experiments showing that FAXs perform well along these metrics as well as being more faithful to the image classifiers than conventional, non-argumentative explanation methods.
-
Conference paperGorur D, Rago A, Toni F, 2025,
Can Large Language Models perform Relation-based Argument Mining?
, The 31st International Conference on Computational Linguistics -
Conference paperRapberger A, Ulbricht M, Toni F, 2024,
On the correspondence of non-flat assumption-based argumentation and logic programming with negation as failure in the head
, 22nd International Workshop on Nonmonotonic Reasoning NMR 24), Publisher: CEUR Workshop Proceedings, Pages: 122-121, ISSN: 1613-0073The relation between (a fragment of) assumption-based argumentation (ABA) and logic programs (LPs) under stable model semantics is well-studied. However, for obtaining this relation, the ABA framework needs to be restricted to being flat, i.e., a fragment where the (defeasible) assumptions can never be entailed, only assumed to be true or false. Here, we remove this restriction and show a correspondence between non-flat ABA and LPs with negation as failure in their head. We then extend this result to so-called set-stable ABA semantics, originally defined for the fragment of non-flat ABA called bipolar ABA. We showcase how to define set-stable semantics for LPs with negation as failure in their head and show the correspondence to set-stable ABA semantics.
-
Conference paperVasileiou S, Kumar A, Yeoh W, et al., 2024,
Dialectical reconciliation via structured argumentative dialogues
, KR 2024We present a novel framework designed to extend model reconciliation approaches, commonly used in human-aware planning, for enhanced human-AI interaction. By adopting a structured argumentation-based dialogue paradigm, our framework enables dialectical reconciliation to address knowledge discrepancies between an explainer (AI agent) and an explainee (human user), where the goal is for the explainee to understand the explainer's decision. We formally describe the operational semantics of our proposed framework, providing theoretical guarantees. We then evaluate the framework's efficacy ``in the wild'' via computational and human-subject experiments. Our findings suggest that our framework offers a promising direction for fostering effective human-AI interactions in domains where explainability is important.
This data is extracted from the Web of Science and reproduced under a licence from Thomson Reuters. You may not copy or re-distribute this data in whole or in part without the written consent of the Science business of Thomson Reuters.
Contact us
Artificial Intelligence Network
South Kensington Campus
Imperial College London
SW7 2AZ
To reach the elected speaker of the network, Dr Rossella Arcucci, please contact:
To reach the network manager, Diana O'Malley - including to join the network - please contact: