Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Video

Updated: May 30, 2026

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
04:48

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography

Published on: November 30, 2022

Loss masking-based gradient optimisation: A new approach for training supervised biomedical named entity recognition

Joshy Alphonse1, Pradeesh Babu1, Srisairam Achutan2

  • 1School of Biotechnology, Amrita Vishwa Vidyapeetham University, Kollam, India.

Scientific Reports
|May 18, 2026
PubMed
Summary

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Epigenetic Reprogramming by <i>Mycobacterium tuberculosis</i> Secretory Proteins: Implications for Pathogenesis and Therapy.

Antibiotics (Basel, Switzerland)·2026
Same author

Deciphering Molecular and Solvent Effects on Aqueous and Organic Solubility through Interpretable Machine Learning Approaches.

ACS omega·2026
Same author

Decoding Host-Pathogen Dynamics in <i>Klebsiella pneumoniae</i> Infections: Mechanisms of Cell Death Regulation and Host-Directed Therapies for Sepsis.

Antibiotics (Basel, Switzerland)·2026
Same author

Synthesis and Evaluation of Novel 2-((1<i>H</i>-1,2,4-triazol-5-yl)thio)-<i>N</i>-benzylidene-<i>N</i>-phenylacetohydrazide as Potential Antimicrobial Agents.

International journal of molecular sciences·2025
Same author

Host-directed insights into mycobacterial infections: Toward targeted immunomodulation.

Microbial pathogenesis·2025
Same author

Repurposing SARS CoV 2 ligands identifies salvianolic acid C as potential antiviral candidate against human metapneumovirus through computational study.

Scientific reports·2025

A new Loss-Masking Optimisation framework enables unified training for Biomedical Named Entity Recognition (BioNER) models across multiple datasets. This approach improves performance by reducing cross-dataset interference, offering a scalable solution for biomedical literature mining.

Area of Science:

  • Computational Biology
  • Natural Language Processing
  • Bioinformatics

Background:

  • The growing volume of biomedical literature requires efficient text mining tools.
  • Biomedical Named Entity Recognition (BioNER) is crucial for extracting information.
  • Current BioNER methods struggle with dataset heterogeneity and scalability.

Purpose of the Study:

  • To develop a unified BioNER model trained across multiple datasets without architectural complexity.
  • To introduce a novel Loss-Masking Optimisation framework for scalable multi-dataset BioNER training.
  • To address the limitations of existing multi-task and collaborative learning approaches.

Main Methods:

  • Proposed a novel Loss-Masking Optimisation framework for BioNER.
  • Implemented a dataset-aware masking strategy to nullify irrelevant tags during training.

Related Experiment Videos

Last Updated: May 30, 2026

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
04:48

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography

Published on: November 30, 2022

  • Extended the standard BERT-based NER pipeline with a tag-masking array.
  • Trained a single BioNER model across 16 diverse biomedical NER datasets.

    • Main Results:

    • Achieved higher precision and overall F1 scores compared to conventional multi-dataset training.
    • Demonstrated improved performance on some datasets, while others remained near baseline or slightly declined.
    • Showcased the effectiveness of the unified model in reducing cross-dataset interference.
    • Validated the scalability of the proposed framework.

    Conclusions:

    • The Loss-Masking Optimisation framework provides a scalable and effective method for unified multi-dataset BioNER training.
    • This approach offers a promising alternative to complex multi-task architectures for biomedical text mining.
    • Dataset interactions have a nuanced impact on model performance, requiring careful consideration.