arXiv subjects

Fariba Afrin Irany

Publications and source records attributed to Fariba Afrin Irany.

5 recordsLinked to original sources

arXiv2026

A Hybrid Retrieval and Reranking Framework for Evidence-Grounded Retrieval-Augmented Generation

Retrieval-augmented generation (RAG) improves large language model reliability by grounding generated responses in external evidence. However, RAG performance depends on the relevance of retrieved passages, the quality of evidence ranking, and the ability to verify whether generated claims are supported by source documents. This study presents a hybrid retrieval and reranking framework for citation-aware RAG in biomedical and healthcare-related document question answering. The framework uses Amazon Bedrock Knowledge Bases for document ingestion, parsing, chunking, embedding generation, and evidence retrieval. Source PDF documents are stored in Amazon S3, embedded using Amazon Titan Text Embeddings V2, and indexed with Amazon OpenSearch Serverless. Hybrid retrieval first retrieves candidate evidence chunks, and Cohere reranking then prioritizes the most relevant passages before answer generation. The answer-generation stage uses top-ranked evidence chunks to produce controlled, evidence-grounded responses, while a separate judge model evaluates each generated factual claim against the retrieved evidence. The framework was evaluated using 25 biomedical NLP and healthcare transformer queries as a pilot-scale proof-of-concept study. Across the evaluation set, the system retrieved and reranked 500 evidence chunks and generated answers from top-ranked evidence. Claim-level grounding evaluation extracted 200 factual claims, all of which were judged to be supported by retrieved evidence, resulting in 100.0% grounding accuracy. The results suggest that hybrid retrieval, reranking, conservative prompting, and claim-level evaluation can support reliable evidence-grounded RAG responses when sufficient source evidence is available.

cs.IR↗

arXiv2026

Selective Fine-Tuning of GPT Architectures for Parameter-Efficient Clinical Text Classification

The rapid expansion of electronic health record (EHR) systems has generated large volumes of unstructured clinical narratives that contain valuable information for disease identification, patient cohort discovery, and clinical decision support. Extracting structured knowledge from these free-text documents remains challenging because clinical language is highly specialized, labeled datasets are limited, and full fine-tuning of large pretrained language models can require substantial computational resources. Efficient adaptation strategies are therefore essential for practical clinical natural language processing applications. This study proposes a parameter-efficient selective fine-tuning framework for adapting GPT-2 to clinical text classification tasks. Instead of updating the entire pretrained model, the majority of network parameters are frozen, and only the final Transformer block, the final layer normalization module, and a lightweight classification head are updated during training. This design substantially reduces the number of trainable parameters while preserving the contextual representation capabilities learned during pretraining. The proposed approach is evaluated using radiology reports from the MIMIC-IV-Note dataset with automatically derived CheXpert-style labels. Experiments on 50,000 radiology reports demonstrate that selective fine-tuning achieves approximately 91% classification accuracy while updating fewer than 6% of the model parameters. Comparative experiments with head-only training and full-model fine-tuning show that the proposed method provides a favorable balance between predictive performance and computational efficiency. These results indicate that selective fine-tuning offers an efficient and scalable framework for clinical text classification.

cs.CL↗

arXiv2026

From Generative Modeling to Clinical Classification: A GPT-Based Architecture for EHR Notes

The increasing availability of unstructured clinical narratives in electronic health records (EHRs) has created new opportunities for automated disease characterization, cohort identification, and clinical decision support. However, modeling long, domain-specific clinical text remains challenging due to limited labeled data, severe class imbalance, and the high computational cost of adapting large pretrained language models. This study presents a GPT-based architecture for clinical text classification that adapts a pretrained decoder-only Transformer using a selective fine-tuning strategy. Rather than updating all model parameters, the majority of the GPT-2 backbone is frozen, and training is restricted to the final Transformer block, the final layer normalization, and a lightweight classification head. This approach substantially reduces the number of trainable parameters while preserving the representational capacity required to model complex clinical language. The proposed method is evaluated on radiology reports from the MIMIC-IV-Note dataset using uncertainty-aware CheXpert-style labels derived directly from report text. Experiments cover multiple problem formulations, including multi-label classification of radiographic findings, binary per-label classification under different uncertainty assumptions, and aggregate disease outcome prediction. Across varying dataset sizes, the model exhibits stable convergence behavior and strong classification performance, particularly in settings dominated by non-mention and negated findings. Overall, the results indicate that selective fine-tuning of pretrained generative language models provides an efficient and effective pathway for clinical text classification, enabling scalable adaptation to real-world EHR data while significantly reducing computational complexity.

cs.CL↗

arXiv2024

Sampling-Based Attack for Centrality Disruption in Complex Networks

Many mobile networks are represented as graphs to obtain insight to their connectivity and transmission properties. Among these properties centrality resilience, that is, how well centralities, such as closeness and betweennesss, are maintained under attacks is a critical factor for proper functioning of a network. In this paper, we study the centrality resilience of complex networks by developing attack models to disrupt the rank of the top path-based centrality vertices. To develop our attack models, we extend the concept of rich clubs of influential vertices to the more general framework of scattered rich clubs. We define scattered rich clubs as dense subgraphs of high centrality vertices that are spread (scattered) across the network. Finding scattered rich clubs, although of polynomial time complexity, is extremely expensive computationally. We use snowball sampling to identify these important substructures as well as to identify which edges to target in our proposed attack models. Our results over a set of real world networks demonstrate that our proposed algorithm is effective in finding the single or scattered rich clubs efficiently and in successfully disrupting the centrality rankings of the network. To summarize, we propose sampling-based attack models for testing the resilience of networks with respect to centrality rankings. As part of this process, we introduce scattered rich clubs, a generalized form of the rich club model, efficient algorithms to detect them, and demonstrate their relation to network resilience.

cs.SI↗

arXiv2021

The Multi-phase spatial meta-heuristic algorithm for public health emergency transportation

The delivery of Medical Countermeasures(MCMs) for mass prophylaxis in the case of a bio-terrorist attack is an active research topic that has interested the research community over the past decades. The objective of this study is to design an efficient algorithm for the Receive Reload and Store Problem(RSS) in which we aim to find feasible routes to deliver MCMs to a target population considering time, physical, and human resources, and capacity limitations. For doing this, we adapt the p-median problem to the POD-based emergency response planning procedures and propose an efficient algorithm solution to perform the p-median in reasonable computational time. We present RE-PLAN, the Response PLan Analyzer system that contains some RSS solutions developed at The Center for Computational Epidemiology and Response Analysis (CeCERA) at the University of North Texas. Finally, we analyze a study case where we show how the computational performance of the algorithm can impact the process of decision making and emergency planning in the short and long terms.

cs.AI↗