Fatema Khan | Searcharxiv

arXiv2025

Bangla MedER: Multi-BERT Ensemble Approach for the Recognition of Bangla Medical Entity

Medical Entity Recognition (MedER) is an essential NLP task for extracting meaningful entities from the medical corpus. Nowadays, MedER-based research outcomes can remarkably contribute to the development of automated systems in the medical sector, ultimately enhancing patient care and outcomes. While extensive research has been conducted on MedER in English, low-resource languages like Bangla remain underexplored. Our work aims to bridge this gap. For Bangla medical entity recognition, this study first examined a number of transformer models, including BERT, DistilBERT, ELECTRA, and RoBERTa. We also propose a novel Multi-BERT Ensemble approach that outperformed all baseline models with the highest accuracy of 89.58%. Notably, it provides an 11.80% accuracy improvement over the single-layer BERT model, demonstrating its effectiveness for this task. A major challenge in MedER for low-resource languages is the lack of annotated datasets. To address this issue, we developed a high-quality dataset tailored for the Bangla MedER task. The dataset was used to evaluate the effectiveness of our model through multiple performance metrics, demonstrating its robustness and applicability. Our findings highlight the potential of Multi-BERT Ensemble models in improving MedER for Bangla and set the foundation for further advancements in low-resource medical NLP.

cs.CL↗

arXiv2021

A Tutorial on Trace-based Simulations of Mobile Ad-hoc Networks on the Example of Aeronautical Communications

The OMNeT++ simulator is well-suited for the simulation of randomized user behavior in communication networks. However, there are scenarios, where such a random model is unsuited to evaluate a communication system, and this paper attempts to highlight such a case. Using this example of ad-hoc communication between aircraft mid-flight, a tutorial-style description is attempted that shall show how the OMNeT++ simulator can be used when a wealth of real-world trace data is available. In particular, it is described how mobility trace files can be directly used within OMNeT++, and how to link the generation of data messages to this mobility data. This is explained via an example simulation that evaluates a communication network in which an aircraft notifies the ground control when it enters or leaves a specific geographic region. Additionally, a novel trace-based application has been developed to achieve this link between mobility and message generation. Furthermore, a new TDMA-based medium access protocol for decentralized communication networks is presented, which is oracle-based and thus allows a TDMA-like behavior of medium access without causing any overhead; it can be useful when upper-layer protocols should be evaluated under the assumption of TDMA-like behavior, but isolated from the effects of a full-fledged TDMA protocol. Finally, physical layer behavior is often either overly simplistic or overly computationally expensive. For the latter case, when a detailed channel model is available but its evaluation requires prohibitive computational effort, then averaging its behavior into trace data can find a middle ground between efficient evaluation and realistic representation. Hence, a novel trace-based radio model has been developed that makes use of an SNR to PER mapping. In the spirit of open science, all implementations have been made available under open licenses.

cs.NI↗