arXiv subjects

Saeed Ahmadi

Publications and source records attributed to Saeed Ahmadi.

7 recordsLinked to original sources

arXiv2026

Perturbation-based Regional Interpretability through Subtraction Mapping (PRISM): naming-error dissociations in language models and post-stroke aphasia

Mechanistic interpretability of large language models lacks spatially resolved, falsifiable tools for testing whether internal components are specialized for distinct cognitive operations. We adapt subtraction analysis, the standard framework of human neuroimaging, from biological brains to perturbed transformers, and apply the same logic to both substrates in parallel. Building on the Brain-LLM Unified Model (BLUM), which showed that layer-perturbed LLaVA-1.6-Vicuna-13B error profiles match the lesion patterns of aphasic patients, we develop PRISM (Perturbation-based Regional Interpretability through Subtraction Mapping). PRISM maps the seven clinical Philadelphia Naming Test categories, subtracts error classes pairwise, and treats each perturbation seed as a subject in a group analysis with threshold-free cluster enhancement along the layer axis. We run a structurally matched analysis on 213 chronic post-stroke aphasia patients using correlation-difference lesion-symptom mapping, and replicate both sides on held-out splits. The designs match in subject dimension (seeds, patients), spatial dimension (layers, atlas-parcellated cortex) and thresholding, but the contrast operator differs: a within-subject error-proportion difference for the LLM, a between-subject correlation difference for the cortex. Both substrates recover a robust phonemic-favoring dissociation, a deep layer cluster and a frontal-perisylvian cortical cluster, both replicating; the semantic-favoring direction is a consistently signed but non-significant trend on both. PRISM thus gives a falsifiable, spatially resolved test of functional-specialization claims in transformer language models. A confirmatory ROI-level intervention (PRISM Stage 3) licensing the strongest causal-mechanism claim is left to subsequent work.

cs.LG↗

arXiv2026

Recovering Lesion Parameters from Aphasic Picture Naming Error Profiles in Large Language Models

Interpretability methods for large language models (LLMs) describe internal state but do not directly test whether that state is causally sufficient to produce the observed behavior. In earlier work, we lesioned LLMs to produce error profiles in picture naming, a central task for assessing aphasia, and found that specific lesions produced errors resembling those of individual stroke survivors. Here we ask the inverse question: given an error profile, can the lesion parameters that produced it be recovered, and what does this inverse problem reveal about transformer computation? Lesions in LLaVA-Vicuna 13B were parameterized by layer index, modification percentage, and noise sigma across 4,840 configurations, and error profiles were characterized by a seven-category clinical taxonomy (correct, semantic, unrelated, formal, mixed, neologism, no-response). We trained a multi-task neural network to map error profiles back to perturbation parameters. The problem admitted a partial solution: across 10 independently trained inverse models, modification percentage and noise sigma were recoverable, whereas layer index was recoverable only within a neighborhood. In counterfactual validation, a fresh model instance perturbed with the recovered parameters reproduced the target behavior in 81.4% of cases. This dissociation between low layer recovery and high counterfactual fidelity is consistent with functional redundancy across transformer layers, a property not captured by standard interpretability methods. As an out-of-distribution test, we applied the trained model to picture-naming error profiles from 278 stroke survivors; recovered parameters were syndrome-discriminative, most strongly for perturbation intensity, indicating generalization beyond the training distribution. Counterfactual validation provides a general framework for LLM interpretability claims beyond inverse mapping.

cs.CL↗

arXiv2026

Lesioned Multimodal Language Models Reproduce Aphasic Picture-Naming Patterns

Aphasia following stroke commonly produces systematic naming errors with characteristic profiles, but whether general-purpose language models not designed for clinical simulation can reproduce these patterns remains untested. We investigated (1) whether lesions or controlled perturbations to a multimodal language model can reproduce different types of errors in picture naming, and (2) whether the framework can reproduce the complete error profile of individual persons with aphasia (PWAs). Using LLaVA 1.6, we evaluated perturbation configurations that varied the layer, proportion, and amount of noise applied to model units. We examined 278 PWAs on the Philadelphia Naming Test, classifying responses into seven categories using a validated neural classifier. Six of seven response categories (correct, semantic, mixed, unrelated, neologism, no response errors) emerged at clinically-comparable proportions across distinct parameter space regions, with formal paraphasia being the exception. Searching the perturbation space revealed configurations that reproduced the individual error profile in at least six of seven categories for 97.8% of PWAs and in all seven categories for 79.5% of PWAs. Monte Carlo baselines confirmed that this matching reflects joint inter-category structure rather than marginal overlap. These results establish a quantitative framework for reproducing individual aphasic error patterns in picture naming. They suggest the potential for language models to serve as digital twins of individuals with post-stroke aphasia.

cs.AI↗

arXiv2026

Stroke Lesions as a Rosetta Stone for Language Model Interpretability

Large language models (LLMs) have achieved remarkable capabilities, yet methods to verify which model components are truly necessary for language function remain limited. Current interpretability approaches rely on internal metrics and lack external validation. Here we present the Brain-LLM Unified Model (BLUM), a framework that leverages lesion-symptom mapping, the gold standard for establishing causal brain-behavior relationships for over a century, as an external reference structure for evaluating LLM perturbation effects. Using data from individuals with chronic post-stroke aphasia (N = 410), we trained symptom-to-lesion models that predict brain damage location from behavioral error profiles, applied systematic perturbations to transformer layers, administered identical clinical assessments to perturbed LLMs and human patients, and projected LLM error profiles into human lesion space. LLM error profiles were sufficiently similar to human error profiles that predicted lesions corresponded to actual lesions in error-matched humans above chance in 67% of picture naming conditions (p < 10^{-23}) and 68.3% of sentence completion conditions (p < 10^{-61}), with semantic-dominant errors mapping onto ventral-stream lesion patterns and phonemic-dominant errors onto dorsal-stream patterns. These findings open a new methodological avenue for LLM interpretability in which clinical neuroscience provides external validation, establishing human lesion-symptom mapping as a reference framework for evaluating artificial language systems and motivating direct investigation of whether behavioral alignment reflects shared computational principles.

cs.LG↗

arXiv2024

Security and Privacy Enhancing in Blockchain-based IoT Environments via Anonym Auditing

The integration of blockchain technology in Internet of Things (IoT) environments is a revolutionary step towards ensuring robust security and enhanced privacy. This paper delves into the unique challenges and solutions associated with securing blockchain-based IoT systems, with a specific focus on anonymous auditing to reinforce privacy and security. We propose a novel framework that combines the decentralized nature of blockchain with advanced security protocols tailored for IoT contexts. Central to our approach is the implementation of anonymization techniques in auditing processes, ensuring user privacy while maintaining the integrity and transparency of blockchain transactions. We outline the architecture of blockchain in IoT environments, emphasizing the workflow and specific security mechanisms employed. Additionally, we introduce a security protocol that integrates privacy-enhancing tools and anonymous auditing methods, including the use of advanced cryptographic techniques for anonymity. This study also includes a comparative analysis of our proposed framework against existing models in the domain. Our work aims to provide a comprehensive blueprint for enhancing security and privacy in blockchain-based IoT environments, paving the way for more secure and private digital ecosystems.

cs.CR↗

arXiv2023

FIDS: Fuzzy Intrusion Detection System for simultaneous detection of DoS/DDoS attacks in Cloud computing

In recent times, I've encountered a principle known as cloud computing, a model that simplifies user access to data and computing power on a demand basis. The main objective of cloud computing is to accommodate users' growing needs by decreasing dependence on human resources, minimizing expenses, and enhancing the speed of data access. Nevertheless, preserving security and privacy in cloud computing systems pose notable challenges. This issue arises because these systems have a distributed structure, which is susceptible to unsanctioned access - a fundamental problem. In the context of cloud computing, the provision of services on demand makes them targets for common assaults like Denial of Service (DoS) attacks, which include Economic Denial of Sustainability (EDoS) and Distributed Denial of Service (DDoS). These onslaughts can be classified into three categories: bandwidth consumption attacks, specific application attacks, and connection layer attacks. Most of the studies conducted in this arena have concentrated on a singular type of attack, with the concurrent detection of multiple DoS attacks often overlooked. This article proposes a suitable method to identify four types of assaults: HTTP, Database, TCP SYN, and DNS Flood. The aim is to present a universal algorithm that performs effectively in detecting all four attacks instead of using separate algorithms for each one. In this technique, seventeen server parameters like memory usage, CPU usage, and input/output counts are extracted and monitored for changes, identifying the failure point using the CUSUM algorithm to calculate the likelihood of each attack. Subsequently, a fuzzy neural network is employed to determine the occurrence of an attack. When compared to the Snort software, the proposed method's results show a significant improvement in the average detection rate, jumping from 57% to 95%.

cs.CR↗

arXiv2022

Privacy-Preserving Cloud Computing: Ecosystem, Life Cycle, Layered Architecture and Future Roadmap

Privacy-Preserving Cloud Computing is an emerging technology with many applications in various fields. Cloud computing is important because it allows for scalability, adaptability, and improved security. Likewise, privacy in cloud computing is important because it ensures that the integrity of data stored on the cloud maintains intact. This survey paper on privacy-preserving cloud computing can help pave the way for future research in related areas. This paper helps to identify existing trends by establishing a layered architecture along with a life cycle and an ecosystem for privacy-preserving cloud systems in addition to identifying the existing trends in research on this area.

cs.CR↗