Preventing Error Propagation in Multi-Agent AI through Runtime Monitoring

Multi-agent AI systems can improve answer selection by allowing different language models to exchange reasoning traces, revise initial predictions, and support a final decision. However, such communication may also introduce reliability risks: reasoning from one agent can correct another agent's mistake, but it can also mislead an agent that was initially correct. This paper studies reliable multi-agent AI communication through reasoning exchange and runtime answer revision. We develop a framework in which agents first answer multiple-choice questions independently, then share reasoning traces and revise their decisions. We conduct numerical experiments where we evaluate whether this process improves accuracy, produces more positive than negative answer transitions, and remains effective across domains such as cybersecurity, networking, and general knowledge. The results help identify when multi-agent reasoning improves reliability and when it may propagate errors.

Memory as an Attack Surface in LLM Agents: A Study on Multiple-Choice Question Answering

AI agents extend conventional large language model (LLM) applications by integrating language understanding with task execution, external tool use, and memory mechanisms. While memory allows agents to retain prior interactions and provide more personalized and context-aware responses, it also introduces a new vulnerability: information stored in memory can influence future outputs even when the current query is clean. In this paper, we investigate memory manipulation in LLM-based agents for multiple-choice question answering. We first design and implement an LLM-based AI agent with an external memory component that stores and retrieves task-relevant information. We then introduce basic memory manipulation scenarios in which misleading or corrupted memories are inserted into the agent before it answers multiple-choice questions. Using a controlled experimental setup, we compare the agent's performance before and after memory manipulation and measure changes in answer accuracy, attack success rate, and selection of manipulated options. Our results show that even simple memory manipulations can noticeably affect the agent's final answers, causing it to select incorrect options despite receiving clean and well-formed questions.

Classification of Single and Mixed Partial Discharges under Switching Voltage Using an AWA-CNN Framework

The growing use of fast-switching power electronics has made partial discharge (PD) analysis under switching-voltage excitation increasingly important, yet more challenging than under sinusoidal conditions due to activity concentrated at voltage transitions. This work presents an Amplitude-Width-Area (AWA) pattern representation for source-oriented PD analysis under switching-voltage excitation. In the proposed method, time domain PD pulses are characterized using pulse amplitude, width, and area, and mapped into a visual pattern where amplitude and area define the coordinate axes and width is encoded by color. The generated AWA patterns are used to distinguish six single and mixed PD source conditions: corona, internal, surface, corona+internal, corona+surface, and internal+surface. To evaluate the classification capability of the proposed representation, a Random Forest baseline and two Convolutional Neural Network (CNN) models, InceptionV3 and ResNet-18, are compared. The AWA patterns show distinguishable source-dependent distributions, and CNN-based classification achieves testing accuracy above 96%, compared with 73.33% for Random Forest. The results indicate that AWA patterns provide a visual representation of PD pulses suitable for multi-class PD source classification under switching-voltage excitation.

Trustworthy Medical Imaging with Large Language Models: A Study of Hallucinations Across Modalities

Large Language Models (LLMs) are increasingly applied to medical imaging tasks, including image interpretation and synthetic image generation. However, these models often produce hallucinations, which are confident but incorrect outputs that can mislead clinical decisions. This study examines hallucinations in two directions: image to text, where LLMs generate reports from X-ray, CT, or MRI scans, and text to image, where models create medical images from clinical prompts. We analyze errors such as factual inconsistencies and anatomical inaccuracies, evaluating outputs using expert informed criteria across imaging modalities. Our findings reveal common patterns of hallucination in both interpretive and generative tasks, with implications for clinical reliability. We also discuss factors contributing to these failures, including model architecture and training data. By systematically studying both image understanding and generation, this work provides insights into improving the safety and trustworthiness of LLM driven medical imaging systems.

Can Large Language Models Challenge CNNS in Medical Image Analysis?

This study presents a multimodal AI framework designed for precisely classifying medical diagnostic images. Utilizing publicly available datasets, the proposed system compares the strengths of convolutional neural networks (CNNs) and different large language models (LLMs). This in-depth comparative analysis highlights key differences in diagnostic performance, execution efficiency, and environmental impacts. Model evaluation was based on accuracy, F1-score, average execution time, average energy consumption, and estimated $CO_2$ emission. The findings indicate that although CNN-based models can outperform various multimodal techniques that incorporate both images and contextual information, applying additional filtering on top of LLMs can lead to substantial performance gains. These findings highlight the transformative potential of multimodal AI systems to enhance the reliability, efficiency, and scalability of medical diagnostics in clinical settings.

Hallucinations and Key Information Extraction in Medical Texts: A Comprehensive Assessment of Open-Source Large Language Models

Clinical summarization is crucial in healthcare as it distills complex medical data into digestible information, enhancing patient understanding and care management. Large language models (LLMs) have shown significant potential in automating and improving the accuracy of such summarizations due to their advanced natural language understanding capabilities. These models are particularly applicable in the context of summarizing medical/clinical texts, where precise and concise information transfer is essential. In this paper, we investigate the effectiveness of open-source LLMs in extracting key events from discharge reports, such as reasons for hospital admission, significant in-hospital events, and critical follow-up actions. In addition, we also assess the prevalence of various types of hallucinations in the summaries produced by these models. Detecting hallucinations is vital as it directly influences the reliability of the information, potentially affecting patient care and treatment outcomes. We conduct comprehensive numerical simulations to rigorously evaluate the performance of these models, further probing the accuracy and fidelity of the extracted content in clinical summarization.

Learning-Based Two-Way Communications: Algorithmic Framework and Comparative Analysis

Machine learning (ML)-based feedback channel coding has garnered significant research interest in the past few years. However, there has been limited research exploring ML approaches in the so-called "two-way" setting where two users jointly encode messages and feedback for each other over a shared channel. In this work, we present a general architecture for ML-based two-way feedback coding, and show how several popular one-way schemes can be converted to the two-way setting through our framework. We compare such schemes against their one-way counterparts, revealing error-rate benefits of ML-based two-way coding in certain signal-to-noise ratio (SNR) regimes. We then analyze the tradeoffs between error performance and computational overhead for three state-of-the-art neural network coding models instantiated in the two-way paradigm.

Multi-Agent Reinforcement Learning for Graph Discovery in D2D-Enabled Federated Learning

Augmenting federated learning (FL) with device-to-device (D2D) communications can help improve convergence speed and reduce model bias through local information exchange. However, data privacy concerns, trust constraints between devices, and unreliable wireless channels each pose challenges in finding an effective yet resource efficient D2D graph structure. In this paper, we develop a decentralized reinforcement learning (RL) method for D2D graph discovery that promotes communication of impactful datapoints over reliable links for multiple learning paradigms, while following both data and device-specific trust constraints. An independent RL agent at each device trains a policy to predict the impact of incoming links in a decentralized manner without exposure of local data or significant communication overhead. For supervised settings, the D2D graph aims to improve device-specific label diversity without compromising system-level performance. For semi-supervised settings, we enable this by incorporating distributed label propagation. For unsupervised settings, we develop a variation-based diversity metric which estimates data diversity in terms of occupied latent space. Numerical experiments on five widely used datasets confirm that the data diversity improvements induced by our method increase convergence speed by up to 3 times while reducing energy consumption by up to 5 times. They also show that our method is resilient to stragglers and changes in the aggregation interval. Finally, we show that our method offers scalability benefits for larger system sizes without increases in relative overhead, and adaptability to various downstream FL architectures and to dynamic wireless environments.

Cooperative Decentralized Backdoor Attacks on Vertical Federated Learning

Federated learning (FL) is vulnerable to backdoor attacks, where adversaries alter model behavior on target classification labels by embedding triggers into data samples. While these attacks have received considerable attention in horizontal FL, they are less understood for vertical FL (VFL), where devices hold different features of the samples, and only the server holds the labels. In this work, we propose a novel backdoor attack on VFL which (i) does not rely on gradient information from the server and (ii) considers potential collusion among multiple adversaries for sample selection and trigger embedding. Our label inference model augments variational autoencoders with metric learning, which adversaries can train locally. A consensus process over the adversary graph topology determines which datapoints to poison. We further propose methods for trigger splitting across the adversaries, with an intensity-based implantation scheme skewing the server towards the trigger. Our convergence analysis reveals the impact of backdoor perturbations on VFL indicated by a stationarity gap for the trained model, which we verify empirically as well. We conduct experiments comparing our attack with recent backdoor VFL approaches, finding that ours obtains significantly higher success rates for the same main task performance despite not using server information. Additionally, our results verify the impact of collusion on attack performance.

Challenging Fairness: A Comprehensive Exploration of Bias in LLM-Based Recommendations

Large Language Model (LLM)-based recommendation systems provide more comprehensive recommendations than traditional systems by deeply analyzing content and user behavior. However, these systems often exhibit biases, favoring mainstream content while marginalizing non-traditional options due to skewed training data. This study investigates the intricate relationship between bias and LLM-based recommendation systems, with a focus on music, song, and book recommendations across diverse demographic and cultural groups. Through a comprehensive analysis conducted over different LLM-models, this paper evaluates the impact of bias on recommendation outcomes. Our findings reveal that bias is so deeply ingrained within these systems that even a simpler intervention like prompt engineering can significantly reduce bias, underscoring the pervasive nature of the issue. Moreover, factors like intersecting identities and contextual information, such as socioeconomic status, further amplify these biases, demonstrating the complexity and depth of the challenges faced in creating fair recommendations across different groups.