Large language models and retrieval-augmented generation systems treat all knowledge as uniformly persistent, ignoring a well-established property of information: that different types of knowledge expire at fundamentally different rates. This paper introduces the Dynamic Epistemic Decay Framework, a formal multi-dimensional theory that characterizes knowledge validity as a function of five independent decay dimensions: temporal decay (𝜆𝜆𝑡𝑡), paradigm decay (𝜆𝜆𝑝𝑝), uncertainty decay (𝜆𝜆𝑢𝑢), dependency decay (𝜆𝜆𝑑𝑑), and zero decay (𝜆𝜆0). We implement this framework as a four-phase retrieval pipeline and evaluate it on the TempQuestions benchmark (n=1,740) against three baselines: standard cosine similarity, BM25 lexical retrieval, and naive recency ranking. Decay-weighted retrieval achieves 92.1% accuracy versus 13.5% for standard semantic retrieval—a 78.6 percentage point improvement—with zero regressions on stable factual queries. On semantically complex temporal benchmarks where lexical heuristics fail, the framework dominates a more resourced BM25 baseline (90.2% vs 1.6% on date-bounded role queries). Epistemic modulation (Phase 4) and dependency graph reasoning (Phase 3) further demonstrate correct mechanism behavior on specialized benchmarks, validated via proof-of-concept implementation. Unlike temporal KG completion approaches that require structured annotation, and unlike contrastive training approaches to time-sensitive RAG, the decay framework is training-free and operates directly over unstructured text corpora. We argue that the decay framework completes the separation of concerns that RAG began: decoupling not just factual storage from model parameters, but factual currency from both.
Pre-trained language models (PLMs) have shown strong potential in Ethereum account modeling and fraud detection. However, existing approaches often overlook the graph-structured nature of transaction networks. In addition, they struggle with the long-tail distribution of account activity, resulting in anisotropic embedding spaces and poor representation quality for low-frequency accounts. In this paper, we present IGT4ETH, a pre-trained Graph Transformer with an isotropy-enhanced post-processing, which explicitly models transaction topologies and mitigates representational anisotropy for Ethereum account classification. IGT4ETH improves structural representation by incorporating structural centrality and role embeddings into an Edge-augmented Graph Transformer, effectively capturing both topological and interaction patterns in transaction graphs. To further mitigate embedding anisotropy, we systematically evaluate various post-processing techniques. Among them, we adopt the Conceptor Negation (CN) method to softly suppress latent features dominated by high-frequency words via matrix conceptors, alongside a modified Focal-InfoNCE loss to enhance directional uniformity and representation balance. Extensive experiments on four real-world Ethereum account classification tasks, including phishing, exchange, mining, and ICO-wallet classification, demonstrate that IGT4ETH consistently outperforms state-of-the-art PLM-based baselines in terms of classification performance.
Open access
Advanced Graph Neural Networks
Topic Modeling
Artificial Intelligence in Healthcare and Education
Zhaohuang Chen, Zhongqi Fu, Tao Liang, Haidong Ma · 6 authors
Abstract Since the proposal of the blockchain, its application scenarios have been continuously expanded. However, the anonymity feature of the blockchain has hindered market regulation, leading to numerous illegal activities such as phishing fraud, which has now become a serious type of crime. Currently, most phishing fraud detection technologies on blockchain platforms use transaction data to construct basic raw transaction graphs and then use neural network methods to mine key information. This study proposes a graph gated recurrent neural network (GGRNN) model that fully integrates temporal and spatial information, effectively utilizing time-related information in the transaction graph. It first takes an account as the center node to obtain its second-order transaction data and then constructs a dynamic transaction graph (DTG). Subsequently, the DTG is fed to the GGRNN to process the temporal features in a gated recurrent unit (GRU) framework and introduce graph convolutional network (GCN) operations to fully use the node neigh-bourhood topology features, obtain the embedded representation of the graph, and then perform graph classification for phishing node detection. To verify the effectiveness of the proposed model, it was applied to real-world Ethereum transaction datasets. Numerical results show that the proposed GGRNN model significantly outperforms state-of-the-art methods.
We present Viturka, a blockchain architecture that replaces wasteful proof-of-work mining with productive federated learning. The core innovation is Proof of Credibility (PoC): a consensus mechanism where block production probability is determined by accumulated reputation from validated AI contributions rather than computational hash power or financial stake. Viturka leverages recent breakthroughs in Zero-Knowledge Machine Learning (ZKML) to achieve cryptographic verification of model training. Validators generate zero-knowledge proofs attesting to correct training execution, enabling instant on-chain verification without trusted intermediaries or statistical consensus mechanisms. By integrating frameworks like EZKL and Lagrange's DeepProve with GPU-accelerated proving via the Icicle library, validation that previously required hours of recomputation now produces mathematical proofs verifiable in milliseconds. Participants earn credibility by contributing quality training data or validating others' contributions. Only the top 10 highest-credibility validators can participate in validation rounds, with mandatory cooldown periods ensuring rotation. The system uses a temporal commit-reveal scheme for data contributions combined with ZK proofs for validation—fake contributions result in permanent bans, while fraudulent validation is mathematically impossible. This creates infrastructure for training AI models on distributed data without central coordination, with economic incentives aligned toward data quality rather than raw computation. Applications range from commercially valuable use cases like DeFi credit scoring—which could unlock over $100B in overcollateralized capital—to public-good AI for rare diseases, minority languages, and environmental monitoring.
Abstract: This paper introduces Knowledge Tensor Lock (KTL), a novel cognitive-structural authentication framework. Unlike conventional mechanisms (passwords, biometrics), KTL anchors identity in the topology of a user’s private semantic associative network. We formalize cognition as a high-rank tensor and verify identity through an interactive challenge-response reconstruction of subgraph structures. Key Contributions: Formalization of the Knowledge Tensor ($\mathcal{K}$) and its graph projection ($G$). Introduction of the Spectral Sketch ($\mathcal{SS}$) for privacy-preserving structural storage. Analysis of heuristic security against AI-adaptive adversaries and model extraction. A roadmap for integrating Zero-Knowledge Proofs (ZKP) for decentralized identity. Note: This is a stabilized preprint (v1.2) intended for establishing conceptual priority in the fields of AI security and cognitive cryptography.
Background Cross-domain federated learning is an innovative machine learning paradigm that allows data owners from different domains to collaboratively train a shared model while preserving data privacy. However, cross-domain federated learning also faces numerous challenges, such as data and system heterogeneity, client reputation management, and potential threats from malicious attackers. Methods To address these issues, this article proposes a secure cross-domain federated learning scheme based on blockchain fair payment. The proposed scheme effectively evaluates and updates the reputation of each client through a reputation management mechanism and allocates fair rewards based on their contributions. Additionally, the scheme employs advanced cryptographic technologies such as blockchain and zero-knowledge proofs to ensure the security and fairness of data and transactions. A series of experiments are conducted to evaluate the performance and fairness of the proposed scheme on multiple datasets and models, and comparisons are conducted with other mainstream federated learning algorithms. MNIST Dataset is available at: https://www.kaggle.com/datasets/hojjatk/mnist-dataset . Fashion-MNIST Dataset is available at https://github.com/zalandoresearch/fashion-mnist . CIFAR-10 Dataset is available at https://www.cs.toronto.edu/~kriz/cifar.html . Results The experimental results demonstrate that the proposed scheme ensures the performance of federated learning while also maintaining its fairness and security. Specifically, the method achieves a test accuracy of 97% on the MNIST dataset, outperforming Federated Averaging (FedAvg) (95%) and Stochastic Controlled Averaging for Federated Learning (SCAFFOLD) (96%). On the FEMNIST dataset, it attains 89% accuracy. In terms of convergence speed, the proposed optimization-based reputation method converges in 26 rounds, which is faster than baseline methods (28–32 rounds). Under data tampering attacks (50-client scenario), the accuracy drop is less than 3%, showing strong robustness. For fairness, the trust difference and reward difference are reduced to 0.10 and 0.08, respectively. The proposed scheme significantly improves the accuracy, convergence speed, robustness, and fairness of cross-domain federated learning, advancing its practical deployment in real-world scenarios. The experimental data is available at: https://zenodo.org/records/15210778 .
NFT prices are shaped by heterogeneous signals including visual appearance, textual narratives, transaction trajectories, and on-chain interactions, yet existing studies often model these factors in isolation and rarely unify multimodal alignment, temporal non-stationarity, and heterogeneous relational dependencies in a leakage-safe forecasting setting. We propose MM-Temporal-Graph, a cross-modal temporal graph transformer framework for explainable NFT valuation and information-centric risk forecasting. The model encodes image, text, transaction time series, and blockchain behavioral features, constructs a heterogeneous NFT interaction graph (co-transaction, shared creator, wallet relation, and price co-movement), and jointly performs relation-aware graph attention and global temporal–structural transformer reasoning with an adaptive fusion gate. A contrastive multimodal alignment objective improves robustness under market drift, while a risk-aware regularizer and a multi-source risk index enable early warning and interpretable attribution across modalities, time segments, and relational neighborhoods. On MultiNFT-T, MM-Temporal-Graph improves MAE from 0.162 to 0.153 and R2 from 0.823 to 0.841 over the strongest multimodal graph baseline, and achieves 87.4% early risk detection accuracy. These results support accurate, robust, and explainable NFT valuation and proactive risk monitoring in Web3 markets.
Frequent security issues with smart contract vulnerabilities have become a pressing challenge in the industry. Conventional program analysis methods lack flexibility and extensibility, leading to high false positive rates. Deep learning approaches are emerging as a new trend to address this issue. Compared to other neural networks, graph convolutional networks can better capture the structural and logical information of smart contracts. However, existing methods do not fully consider the scale-free characteristics of smart contracts and fail to leverage their complex hierarchical structures and semantic information. Therefore, we develop an end-to-end vulnerability detection framework using Riemannian Graph Convolutional Networks (RGCNet). We first construct smart contract graphs that are rich in semantic and structural information. Next, we learn features of the smart contract graph in the Riemannian manifold, thereby better reflecting its actual topology. Simultaneously, the word embedding network extracts semantic features, forming an end-to-end network where modules promote one another. Extensive experiments are conducted on three vulnerabilities using real-world smart contracts. The results show that the proposed approach exhibits superior performance over state-of-the-art methodologies in terms of accuracy, precision, and recall.
Federated Learning (FL) has emerged as a transformative paradigm for distributed machine learning, enabling model training across decentralized edge devices while preserving data privacy. This methodology is critical for sectors handling sensitive information, such as finance, healthcare, and the Internet of Things (IoT). Despite its benefits, the coordination and communication overhead between distributed nodes remain significant challenges. This paper evaluates the efficacy of REST and GraphQL API architectures in facilitating FL workflows. While REST APIs are favored for their statelessness and simplicity, GraphQL offers enhanced flexibility and efficiency by enabling precise data fetching—a vital feature for bandwidth-constrained decentralized systems. We provide a comparative analysis of these paradigms across performance, security, and scalability metrics, specifically regarding data synchronization and model aggregation. Finally, we propose design best practices for developing APIs that support robust, compliant, and efficient federated prediction systems.
Graph neural networks (GNNs) have shown notable success in identifying security vulnerabilities within Ethereum smart contracts by capturing structural relationships encoded in control- and data-flow graphs. Despite their eff... | Find, read and cite all the research you need on Tech Science Press
The rise of Bitcoin has revolutionized the financial landscape, but it has also opened the door to a new era of criminal activities. Criminals take advantage of the anonymity provided by Bitcoin to conduct illicit transactions and engage in fraudulent activities. To address this issue, this paper proposes a detection model using Graph Neural Networks (GNNs) to detect fraudulent activities in the complex financial systems of Bitcoin. From the GNNs, we use EvolveGCN and EvolveGGCN to compare between them and find a powerful model that can investigate the network construction of financial transactions and capture patterns and anomalies that traditional methods may miss. In the literature, there have been a limited number of studies on Bitcoin fraud detection using GNNs, especially EvolveGGCN. Therefore, in this paper, we focus on the detection of fraud in the Bitcoin network using EvolveGCN and EvolveGGCN. In addition, we used a more recent dataset called Elliptic++, which is an extension of the Elliptic Dataset. The dataset provides valuable information on the behavior and patterns of fraudulent actions in the Bitcoin network. The results show that EvolveGGCN outperforms other models in terms of precision, recall, F1 score, and micro-F1 score. With an F1-score of 0.90 and micro-F1 of 0.93 for detecting illicit transactions in the early time steps.
Wassim Sliti, Félix Cuadrado, Leandro Campos Hernáez, Juan C. Dueñas
Illicit activities and coordinated manipulations in the Non-Fungible Tokens (NFT) market remain significant concerns, driven by the pseudonymous and publicly transparent nature of blockchain transactions and the lack of market oversight. In this study, we introduce a novel framework for detecting suspicious behavior in NFT trading ecosystems through temporal graph analysis. Our approach represents the market as a large-scale, time-evolving transactional graph, capturing realistic market dynamics and the complex interactions between traders.By leveraging temporal graphs, we track not only connections between traders but also the evolution of these interactions over time, including their sequence, rhythm, and frequency, enabling the identification of anomalous behaviors that static graph representations cannot reveal and that may warrant further examination. Using an optimized temporal cycle-detection algorithm, we extract connected groups of wallets for in-depth behavioral analysis, uncovering patterns indicative of unusual coordinated manipulation. To overcome the absence of labeled ground-truth validation data, we employ a temporal motif–based validation, demonstrating that flagged entities exhibit trading behaviors significantly deviating from standard market dynamics. Our results highlight the potential of temporal graph–based methodologies to provide a scalable and effective risk-profiling and market-surveillance tool, assisting analysts and regulatory entities in narrowing the investigation space within the large, permissionless NFT markets and enhancing surveillance, risk detection, and regulatory oversight in decentralized NFT markets.
Alex J. Sinclair, Yuhan Zhang, Michael P. O’Connor, Li Wei
Graph Neural Networks (GNNs) have become the de facto standard for modeling blockchain transaction networks. However, standard GNN architectures predominantly operate under the assumption of homophily-that connected nodes share similar labels or features. This assumption catastrophically fails in the context of Ethereum phishing detection, a quintessential heterophilic learning problem where fraudsters (phishing accounts) actively connect with unlike victims (normal accounts) to facilitate theft. In this work, we argue that the "homophily bottleneck" in existing detection systems obscures critical high-frequency signals necessary for identifying illicit activity. We introduce the Heterophilic Semantic Graph Framework (HSGF), a novel architecture designed to decouple structural roles from feature smoothing. HSGF integrates a Motif-based Semantic Sampling (MSS) strategy to capture complex, directed transactional intents (e.g., dispersing, gathering, mixing) and a Heterophily-Aware Feature Fusion (HAFF) module that prevents the oversmoothing of fraudster representations into victim representations. Extensive experiments on real-world Ethereum datasets demonstrate that HSGF significantly outperforms state-of-the-art baselines, particularly in class-imbalanced scenarios, effectively breaking the ceiling imposed by traditional homophilic aggregation.
Ye Tian, Liangliang Song, Yuanyuan Ma, Yanbin Wang · 7 authors
Detecting fraudulent accounts on Ethereum is critical for securing decentralized finance (DeFi) ecosystems. Although temporal transaction dynamics offer richer behavioral signals than static graphs, existing methods struggle to jointly model the continuous, directional, and community-level nature of fraudulent fund flows. Specifically, they inadequately capture two key cues: directional-temporal transaction patterns (integrating directionality and temporal dynamics) and higher-level behavioral communities formed by accounts with similar transactional and temporal traits.To address these, we propose TimeTrace, an unsupervised graph representation learning framework for fraud detection in Ethereum transaction networks. TimeTrace models temporal behaviors via a Directed Temporal Aggregation mechanism that explicitly distinguishes incoming and outgoing flows while adaptively emphasizing recent interactions to capture pattern evolution. Building on these directional representations, TimeTrace further incorporates a differentiable clustering module to identify latent behavioral communities and encode cluster-level relational semantics. To enhance representation coherence and structural consistency, we introduce Cohesive Embedding Regularization (CER)—a graph refinement objective combining a global Laplacian term for structural consistency, an intra-cluster Laplacian term for cluster compactness, and a fidelity term to retain initial clustering information. Additionally, we construct a new phishing account benchmark with up-to-date Ethereum transaction records for realistic evaluation. Extensive experiments on three datasets demonstrate that TimeTrace outperforms state-of-the-art methods in both accuracy and efficiency, achieving F1-score improvements of 3.62\%-10.83\% and inference speeds at least as fast as the quickest baseline.
With the rapid advancement of Web 3.0 technologies, public blockchain platforms are witnessing the emergence of novel services designed to enhance user privacy and anonymity. However, the powerful untraceability features inherent in these services inadvertently make them attractive tools for criminals seeking to launder illicit funds. Notably, existing de-anonymization methods face three major challenges when dealing with such transactions: highly homogenized transactional semantics, limited ability to model temporal discontinuities, and insufficient consideration of structural sparsity in account association graphs. To address these, we propose GradWATCH, designed to track anonymous accounts in Ethereum privacy-preserving services. Specifically, we first design a learnable account feature mapping module to extract informative transactional semantics from raw on-chain data. We then incorporate transaction relations into the account association graph to alleviate the adverse effects of structural sparsity. To capture temporal evolution, we further propose an edge-aware sliding-window mechanism that propagates and updates gradients at three granularities. Finally, we identify accounts controlled by the same entity by measuring their embedding distances in the learned representation space. Experimental results show that even under the conditions of unbalanced labels and sparse transactions, GradWATCH still achieves significant performance gains, with relative improvements ranging from 1.62% to 15. 22% in the MRR and from 3. 85% to 7. 31% in the F_1.
Non-fungible tokens (NFTs) have become a key asset class in Web3 markets, where visual artwork, textual narratives, and on-chain transaction patterns jointly determine value, yet their pricing dynamics remain volatile, opaque, and difficult to explain. Existing NFT valuation methods typically either ignore the multimodal nature of NFTs or treat assets as independent samples, failing to exploit the rich relational structures induced by shared creators, collections, and ownership patterns, and offering limited interpretability for high-stakes financial decisions. To address these challenges, we propose NFT-Insight, a multimodal graph transformer framework that unifies visual, textual, and blockchain information on a heterogeneous NFT graph and explicitly links structural and content signals to valuation behavior. The framework identifies closely related NFTs via a joint similarity measure in the multimodal embedding space, propagates information through a relation-specific graph attention network and a global transformer encoder, and adopts a regularization strategy that encourages consistent valuations for highly similar assets while still allowing data-driven differentiation. In addition, NFT-Insight integrates attention-based and SHAP-based explanations into a unified analysis pipeline, enabling joint study of valuation behavior and feature attributions at the level of related NFT pairs. Experiments on three large-scale, real-world NFT datasets show that NFT-Insight consistently outperforms strong unimodal, multimodal, and graph-based baselines, reducing MAE and RMSE by up to about 20% in static valuation (withR2up to 0.904), achieving robust cross-market performance with averageR2≈ 0.84 andr≈ 0.93, and attainingR2= 0.911 in temporal forecasting. Interpretability analysis reveals that visual, textual, blockchain, and graph-relational features achieve a high alignment between SHAP importance and attention weights (average Spearman correlation above 0.8), and case studies highlight meaningful valuation patterns driven by rarity, speculative trading, and temporal market shocks. Overall, the proposed framework offers a multimodal graph–based perspective on explainable NFT valuation and market forecasting, and provides a general template for incorporating complex relational and content interactions into graph-based learning in decentralized digital economies.
While Decentralized Autonomous Organizations (DAOs) and Artificial Intelligence are reshaping the governance of academic societies, reliably integrating on-chain decisions with off-chain physical activities remains a critical challenge. The fundamental bottleneck is the difficulty of reliably integrating real-world execution outcomes into the digital decision-making loop. To address this, we propose an endogenous contribution evaluation framework integrating Decentralized Physical Infrastructure Networks (DePIN) and Vision-Language-Action (VLA) models. This approach maps physical entities to on-chain decentralized identities. By leveraging VLA edge nodes to analyze multimodal behavioral data collected via DePIN, the system autonomously generates a verifiable Proof of Real-World Contribution (PoRWC). This proof subsequently drives on-chain incentive distribution through a reputation-weighted consensus mechanism. Consequently, this framework establishes an endogenously trustworthy closed loop from physical processes to digital governance. We demonstrate its feasibility and scalability through a case study of the Chinese Association of Automation (CAA), providing a robust engineering path for the parallel governance of modern academic societies.
Although Large Language Models (LLM) have shown impressive performance across various domains, there is a shortage of benchmarks for systematically evaluating their in-depth understanding of specialized fields such as blockchain. This study extends the Self-Instruct methodology to introduce BLADE (Blockchain Large Language model Assessment Dataset for Evaluation), a comprehensive benchmark dataset for assessing LLM comprehension in the blockchain domain. BLADE consists of a total of 1,382 questions organized according to a systematic classification of blockchain knowledge, featuring a detailed structure with 15 main categories and 5 sub-categories for each. The benchmark covers the entire spectrum of blockchain knowledge, from its fundamentals to consensus mechanisms, architecture, smart contracts, token economy, Decentralized Finance (DeFi), NFT(Non-Fungible Token)s and digital assets, security, governance, and real-world application cases. In this research, we present a benchmark generation methodology utilizing the domain knowledge of GPT-4.5, which allowed us to create high-quality evaluation items of varying difficulty and types from expert-verified seed questions. The evaluation results of various open-source LLMs, including Qwen, DeepSeek, and Kanana, on BLADE showed that current models exhibit significant differences in their understanding of blockchain, with Qwen2.5-7B-Instruct-1M achieving the highest performance. The BLADE benchmark provides a tool for precisely evaluating and improving the blockchain comprehension of LLMs, thereby promoting the effective fusion of AI and blockchain technology and contributing to the development of more reliable decentralized systems.
Smart contracts have been widely applied in various fields. Due to the immuta-bility of data on the blockchain, it is of great significance to conduct smart con-tract vulnerability detection before data is uploaded to the chain. To address the problems of low accuracy and single vulnerability type in traditional detection methods, a blockchain smart contract vulnerability detection method based on Graph Neural Network (GNN) is proposed. This method abstracts the functions and key code segments in smart contracts into nodes in a graph, and constructs edges by leveraging data and control dependencies during code execution, thereby accurately depicting the specific graph structures of reentrancy attacks and timestamp-dependent vulnerabilities. To further enhance the model’s sensi-tivity to key vulnerability patterns, the multi-head attention mechanism is in-novatively introduced, which can effectively screen out the nodes and edges that contribute the most to vulnerability detection, suppress irrelevant or noisy information, and significantly improve the accuracy and robustness of vulnera-bility detection. Experimental results show that the proposed method achieves an accuracy of 85.19% in reentrancy vulnerability detection and 82.37% in timestamp-dependent vulnerability detection, demonstrating excellent vulner-ability identification capability.
The pseudo-anonymity and rapidly expanding ecosystem of Decentralized Finance (DeFi) have brought about significant liquidity on EVM-compatible blockchains, making them lucrative targets for cybercriminals. In the modern financial landscape, the need for an automated, high-speed, and effective illicit money tracing system is more urgent than ever to support regulators, on-chain service providers and security practitioners in their efforts to combat the frequent and large-scale occurrences of cyber financial crimes. In this paper, we propose MFTracer, an automated system for tracing illicit money flows on EVM-compatible blockchains. Against the backdrop of a domain where tracing remains labor-intensive and expert-driven, MFTracer is developed in response to two pressing real-world demands: operational efficiency and forensic effectiveness. In response to the sophisticated fund transfer mechanisms enabled by the EVM environment, we introduce a novel fine-grained technique that enables protocol-agnostic transaction-level fund flow analysis. We further propose MFA, a lightweight and purpose-built graph abstraction with a tailored storage backend, to support efficient data retrieval. We also present a simulation algorithm for downstream illicit flow discovery. We implemented MFTracer. Its infrastructure for data retrieval achieves 3.7× to 9.4× higher storage efficiency while being 14.1× to 300× faster than the leading graph database systems. Furthermore, applied to real-world cybercrime incidents, MFTracer achieved 94.09% coverage of illicit money flows. It also newly reported 686 blockchain addresses and 4183 related transactions involved in money laundering that were previously undiscovered. MFTracer was able to reconstruct complete fund flow trajectories and provide strong evidence to investigators for 120.9 million in stolen assets.
Graph-structured data has become central to modern analytics, enabling institutions to model relationships in domains such as healthcare, finance, cyber security, and education. However, privacy regulations and institutional policies restrict the sharing of sensitive nodes, edges, or interaction logs, preventing the discovery of global graph patterns. This paper introduces a novel framework for Federated Graph Pattern Mining Across Institutions (FGPM-AI), enabling multiple organizations to collaboratively extract global sub graphs, motifs, and temporal patterns without sharing raw graph data. The framework proposes six novel contributions: (1) Privacy-Preserving Pattern Signatures (PPPS) for anonymized sub graph encoding, (2) Federated Temporal Graph Pattern Mining (FT-GPM) to learn evolving patterns across distributed graphs, (3) Zero-Exchange Federated Sub graph Matching (ZE-FSM) using zero-knowledge proofs, (4) Heterogeneity-Aware Graph Pattern Consensus (HGPC) for semantic alignment between distinct graph schemas, (5) Communication-Adaptive Pattern Sharing (CA-FGM) for bandwidth-efficient collaboration, and (6) Multi-Party Graph Pattern Distillation (MGPD) for merging patterns into a unified knowledge model. Experimental design considerations demonstrate the feasibility and robustness of the framework. The results highlight FGPM-AI as a promising direction for secure, scalable, and intelligent cross-institution graph analytics.
Large Language Models (LLMs) are increasingly deployed in high-stakes financial domains, yet they suffer from specific, reproducible hallucinations when performing arithmetic operations. Current mitigation strategies often treat the model as a black box. In this work, we propose a mechanistic approach to intrinsic hallucination detection. By applying Causal Tracing to the GPT-2 XL architecture on the ConvFinQA benchmark, we identify a dual-stage mechanism for arithmetic reasoning: a distributed computational scratchpad in middle layers (L12-L30) and a decisive aggregation circuit in late layers (specifically Layer 46). We verify this mechanism via an ablation study, demonstrating that suppressing Layer 46 reduces the model's confidence in hallucinatory outputs by 81.8%. Furthermore, we demonstrate that a linear probe trained on this layer generalizes to unseen financial topics with 98% accuracy, suggesting a universal geometry of arithmetic deception.
Blockchain technology has emerged as a transformative paradigm for decentralized systems, enabling secure, transparent, and tamper-resistant data management through distributed consensus mechanisms that eliminate the need for centralized control. At the core of these systems, consensus protocols ensure agreement among network participants; however, traditional approaches such as Proof of Work (PoW), Proof of Stake (PoS), and Byzantine Fault Tolerance (BFT) face persistent challenges related to scalability, energy consumption, and latency. In response, graph-theoretic approaches have gained prominence as an effective framework for modeling and optimizing blockchain consensus by representing nodes as vertices and communication links as edges, thereby capturing complex network relationships, trust structures, and interaction patterns. This paper systematically reviews graph-based methods applied to blockchain consensus, highlighting their role in improving efficiency, enhancing security against attacks such as Sybil and double-spending, and optimizing node selection. Advanced techniques including graph partitioning, spectral clustering, and network flow optimization further contribute to improved scalability and throughput. The study identifies a clear transition toward intelligent, hybrid consensus mechanisms integrating graph theory, machine learning, and distributed computing, while also addressing ongoing challenges such as computational complexity and dynamic adaptability, and outlining future directions for AI-driven, scalable, and secure consensus models.