Hallucination Detection in LLMs with Topological Divergence on Attention Graphs

• URL: http://arxiv.org/abs/2504.10063v2
• Date: Thu, 22 May 2025 12:49:21 GMT
• Title: Hallucination Detection in LLMs with Topological Divergence on Attention Graphs
• Authors: Alexandra Bazarova, Aleksandr Yugay, Andrey Shulga, Alina Ermilova, Andrei Volodichev, Konstantin Polev, Julia Belikova, Rauf Parchiev, Dmitry Simakov, Maxim Savchenko, Andrey Savchenko, Serguei Barannikov, Alexey Zaytsev,
• Abstract summary: Hallucination, i.e., generating factually incorrect content, remains a critical challenge for large language models.<n>We introduce TOHA, a TOpology-based HAllucination detector in the RAG setting.
• Score: 64.74977204942199
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Hallucination, i.e., generating factually incorrect content, remains a critical challenge for large language models (LLMs). We introduce TOHA, a TOpology-based HAllucination detector in the RAG setting, which leverages a topological divergence metric to quantify the structural properties of graphs induced by attention matrices. Examining the topological divergence between prompt and response subgraphs reveals consistent patterns: higher divergence values in specific attention heads correlate with hallucinated outputs, independent of the dataset. Extensive experiments - including evaluation on question answering and summarization tasks - show that our approach achieves state-of-the-art or competitive results on several benchmarks while requiring minimal annotated data and computational resources. Our findings suggest that analyzing the topological structure of attention matrices can serve as an efficient and robust indicator of factual reliability in LLMs.

Related papers

• Information-theoretic Quantification of High-order Feature Effects in Classification Problems [0.19791587637442676]
  We present an information-theoretic extension of the High-order interactions for Feature importance (Hi-Fi) method.<n>Our framework decomposes feature contributions into unique, synergistic, and redundant components.<n>Results indicate that the proposed estimator accurately recovers theoretical and expected findings.
  arXiv  Detail & Related papers  (2025-07-06T11:50:30Z)
• How do Large Language Models Understand Relevance? A Mechanistic Interpretability Perspective [64.00022624183781]
  Large language models (LLMs) can assess relevance and support information retrieval (IR) tasks.<n>We investigate how different LLM modules contribute to relevance judgment through the lens of mechanistic interpretability.
  arXiv  Detail & Related papers  (2025-04-10T16:14:55Z)
• Don't Take Things Out of Context: Attention Intervention for Enhancing Chain-of-Thought Reasoning in Large Language Models [32.71672086718058]
  Few-shot Chain-of-Thought (CoT) significantly enhances the reasoning capabilities of large language models (LLMs)<n>We observe that isolated segments, words, or tokens within CoT demonstrations can unexpectedly disrupt the generation process of LLMs.<n>We propose a Few-shot Attention Intervention method (FAI) that dynamically analyzes the attention patterns of demonstrations to accurately identify these tokens.
  arXiv  Detail & Related papers  (2025-03-14T07:46:33Z)
• SINdex: Semantic INconsistency Index for Hallucination Detection in LLMs [2.805517909463769]
  Large language models (LLMs) are increasingly deployed across diverse domains, yet they are prone to generating factually incorrect outputs.<n>We introduce a novel and scalable uncertainty-based semantic clustering framework for automated hallucination detection.
  arXiv  Detail & Related papers  (2025-03-07T23:25:19Z)
• VisFactor: Benchmarking Fundamental Visual Cognition in Multimodal Large Language Models [62.667142971664575]
  We introduce VisFactor, a novel benchmark derived from the Factor-Referenced Cognitive Test (FRCT)<n>VisFactor digitalizes vision-related FRCT subtests to systematically evaluate MLLMs across essential visual cognitive tasks.<n>We present a comprehensive evaluation of state-of-the-art MLLMs, such as GPT-4o, Gemini-Pro, and Qwen-VL.
  arXiv  Detail & Related papers  (2025-02-23T04:21:32Z)
• Understanding Ranking LLMs: A Mechanistic Analysis for Information Retrieval [20.353393773305672]
  We employ a probing-based analysis to examine neuron activations in ranking LLMs.<n>Our study spans a broad range of feature categories, including lexical signals, document structure, query-document interactions, and complex semantic representations.<n>Our findings offer crucial insights for developing more transparent and reliable retrieval systems.
  arXiv  Detail & Related papers  (2024-10-24T08:20:10Z)
• Causally-Aware Unsupervised Feature Selection Learning [15.20376149047008]
  Unsupervised feature selection (UFS) has recently gained attention for its effectiveness in processing unlabeled high-dimensional data.<n>Previous graph-based methods fail to account for the differing impacts of non-causal and causal features in constructing the similarity graph.<n>A novel UFS method, called Causally-Aware UnSupErvised Feature Selection learning (CAUSE-FS), is proposed.
  arXiv  Detail & Related papers  (2024-10-16T04:41:38Z)
• Massive Activations in Graph Neural Networks: Decoding Attention for Domain-Dependent Interpretability [0.9499648210774584]
  We show the emergence of Massive Activations (MAs) within attention layers in edge-featured Graph Neural Networks (GNNs)<n>Our study assesses various edge-featured attention-based GNN models using benchmark datasets, including ZINC, TOX21, and PROTEINS.
  arXiv  Detail & Related papers  (2024-09-05T12:19:07Z)
• Feature graphs for interpretable unsupervised tree ensembles: centrality, interaction, and application in disease subtyping [0.24578723416255746]
  Feature selection assumes a pivotal role in enhancing model interpretability. The accuracy gained from aggregating decision trees comes at the expense of interpretability. The study introduces novel methods to construct feature graphs from unsupervised random forests.
  arXiv  Detail & Related papers  (2024-04-27T12:47:37Z)
• PoLLMgraph: Unraveling Hallucinations in Large Language Models via State Transition Dynamics [51.17512229589]
  PoLLMgraph is a model-based white-box detection and forecasting approach for large language models. We show that hallucination can be effectively detected by analyzing the LLM's internal state transition dynamics. Our work paves a new way for model-based white-box analysis of LLMs, motivating the research community to further explore, understand, and refine the intricate dynamics of LLM behaviors.
  arXiv  Detail & Related papers  (2024-04-06T20:02:20Z)
• Spectral Clustering for Directed Graphs via Likelihood Estimation on Stochastic Block Models [22.421702511126373]
  We leverage statistical inference on block models to guide the development of a spectral clustering algorithm for directed graphs.<n>We establish a theoretical upper bound on the misclustering error of its spectral relaxation, and based on this relaxation, introduce a novel, self-adaptive spectral clustering method for directed graphs.
  arXiv  Detail & Related papers  (2024-03-28T15:47:13Z)
• Zero-shot Causal Graph Extrapolation from Text via LLMs [50.596179963913045]
  We evaluate the ability of large language models (LLMs) to infer causal relations from natural language. LLMs show competitive performance in a benchmark of pairwise relations without needing (explicit) training samples. We extend our approach to extrapolating causal graphs through iterated pairwise queries.
  arXiv  Detail & Related papers  (2023-12-22T13:14:38Z)
• Position-aware Structure Learning for Graph Topology-imbalance by Relieving Under-reaching and Over-squashing [67.83086131278904]
  Topology-imbalance is a graph-specific imbalance problem caused by the uneven topology positions of labeled nodes. We propose a novel position-aware graph structure learning framework named PASTEL. Our key insight is to enhance the connectivity of nodes within the same class for more supervision information.
  arXiv  Detail & Related papers  (2022-08-17T14:04:21Z)
• Heterogeneous Graph Neural Networks using Self-supervised Reciprocally Contrastive Learning [102.9138736545956]
  Heterogeneous graph neural network (HGNN) is a very popular technique for the modeling and analysis of heterogeneous graphs. We develop for the first time a novel and robust heterogeneous graph contrastive learning approach, namely HGCL, which introduces two views on respective guidance of node attributes and graph topologies. In this new approach, we adopt distinct but most suitable attribute and topology fusion mechanisms in the two views, which are conducive to mining relevant information in attributes and topologies separately.
  arXiv  Detail & Related papers  (2022-04-30T12:57:02Z)
• Scalable Intervention Target Estimation in Linear Models [52.60799340056917]
  Current approaches to causal structure learning either work with known intervention targets or use hypothesis testing to discover the unknown intervention targets. This paper proposes a scalable and efficient algorithm that consistently identifies all intervention targets. The proposed algorithm can be used to also update a given observational Markov equivalence class into the interventional Markov equivalence class.
  arXiv  Detail & Related papers  (2021-11-15T03:16:56Z)
• Learning Neural Causal Models with Active Interventions [83.44636110899742]
  We introduce an active intervention-targeting mechanism which enables a quick identification of the underlying causal structure of the data-generating process. Our method significantly reduces the required number of interactions compared with random intervention targeting. We demonstrate superior performance on multiple benchmarks from simulated to real-world data.
  arXiv  Detail & Related papers  (2021-09-06T13:10:37Z)
• Multilingual Multi-Aspect Explainability Analyses on Machine Reading Comprehension Models [76.48370548802464]
  This paper focuses on conducting a series of analytical experiments to examine the relations between the multi-head self-attention and the final MRC system performance. We discover that passage-to-question and passage understanding attentions are the most important ones in the question answering process. Through comprehensive visualizations and case studies, we also observe several general findings on the attention maps, which can be helpful to understand how these models solve the questions.
  arXiv  Detail & Related papers  (2021-08-26T04:23:57Z)
• Counterfactual Maximum Likelihood Estimation for Training Deep Networks [83.44219640437657]
  Deep learning models are prone to learning spurious correlations that should not be learned as predictive clues. We propose a causality-based training framework to reduce the spurious correlations caused by observable confounders. We conduct experiments on two real-world tasks: Natural Language Inference (NLI) and Image Captioning.
  arXiv  Detail & Related papers  (2021-06-07T17:47:16Z)
• Attention improves concentration when learning node embeddings [1.2233362977312945]
  Given nodes labelled with search query text, we want to predict links to related queries that share products. Experiments with a range of deep neural architectures show that simple feedforward networks with an attention mechanism perform best for learning embeddings. We propose an analytically tractable model of query generation, AttEST, that views both products and the query text as vectors embedded in a latent space.
  arXiv  Detail & Related papers  (2020-06-11T21:21:12Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.