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 Concept Videos

You might also read

Related Articles

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

Sort by
Same author

Effect of Aging Treatment on Microstructural Evolution and Mechanical Properties of the Electron Beam Cold Hearth Melting Ti-6Al-4V Alloy.

Materials (Basel, Switzerland)·2022
Same author

Sex-specific associations between legacy and novel per- and polyfluoroalkyl substances and telomere length in newborns in Wuhan, China: Mixture and single pollutant associations.

The Science of the total environment·2022
Same author

Aldosterone-stimulated endothelial epithelial sodium channel (EnNaC) plays a role in cold exposure-induced hypertension in rats.

Frontiers in pharmacology·2022
Same author

Camphene as a Mild, Bio-Derived Porogen for Near-Ambient Processing and 3D Printing of Porous Thermoplastics.

ACS applied materials & interfaces·2022
Same author

Prenatal exposure to perfluorooctane sulfonate alternatives and associations with neonatal thyroid stimulating hormone concentration: A birth cohort study.

Chemosphere·2022
Same author

Autocollimation angle-measurement method with a large range based on spot deformation.

Optics express·2022

Related Experiment Video

Updated: Nov 14, 2025

Decoding Natural Behavior from Neuroethological Embedding
08:00

Decoding Natural Behavior from Neuroethological Embedding

Published on: October 3, 2025

252

Hungarian layer: A novel interpretable neural layer for paraphrase identification.

Han Xiao1

  • 1Artificial Intelligence Department, School of Informatics, Xiamen University, Xiamen, China; Department of Information Communication Technology, Xiamen University Malaysia, Sepang, Malaysia.

Neural Networks : the Official Journal of the International Neural Network Society
|August 18, 2020
PubMed
Summary

This study introduces a novel neural network approach for paraphrase identification, enhancing sequence alignment by incorporating the Hungarian algorithm to better model unmatched sentence parts. The new method significantly improves paraphrase detection accuracy compared to existing techniques.

Keywords:
Hungarian LayerNeural GraphParaphrase Identification

More Related Videos

Author Spotlight: Investigating the Impact of Emotional Prosodies on Voice Recognition and Perception
05:48

Author Spotlight: Investigating the Impact of Emotional Prosodies on Voice Recognition and Perception

Published on: August 9, 2024

1.8K
Dissociation of the Confounding Influences of Expectancy and Integrative Difficulty Residing in Anomalous Sentences in Event-related Potential Studies
05:22

Dissociation of the Confounding Influences of Expectancy and Integrative Difficulty Residing in Anomalous Sentences in Event-related Potential Studies

Published on: May 9, 2019

5.6K

Related Experiment Videos

Last Updated: Nov 14, 2025

Decoding Natural Behavior from Neuroethological Embedding
08:00

Decoding Natural Behavior from Neuroethological Embedding

Published on: October 3, 2025

252
Author Spotlight: Investigating the Impact of Emotional Prosodies on Voice Recognition and Perception
05:48

Author Spotlight: Investigating the Impact of Emotional Prosodies on Voice Recognition and Perception

Published on: August 9, 2024

1.8K
Dissociation of the Confounding Influences of Expectancy and Integrative Difficulty Residing in Anomalous Sentences in Event-related Potential Studies
05:22

Dissociation of the Confounding Influences of Expectancy and Integrative Difficulty Residing in Anomalous Sentences in Event-related Potential Studies

Published on: May 9, 2019

5.6K

Area of Science:

  • Natural Language Processing
  • Computational Linguistics
  • Machine Learning

Background:

  • Paraphrase identification is crucial in natural language processing (NLP).
  • Traditional methods use attention mechanisms for sequence alignment but struggle with unmatched parts.
  • Unmatched aligned parts are vital for accurate paraphrase detection.

Purpose of the Study:

  • To enhance paraphrase identification by improving the modeling of aligned unmatched parts.
  • To integrate the Hungarian algorithm into neural architectures for sequence alignment.
  • To develop a robust method for distinguishing paraphrases using novel alignment techniques.

Main Methods:

  • Encoding sentences into hidden representations using BiLSTM or BERT.
  • Utilizing a Hungarian layer to extract aligned unmatched parts from these representations.
  • Applying cosine similarity to the extracted parts for final paraphrase discrimination.

Main Results:

  • The proposed model significantly outperforms existing baseline methods in paraphrase identification.
  • The integration of the Hungarian algorithm effectively captures crucial unmatched parts.
  • Experimental results demonstrate substantial and significant improvements in accuracy.

Conclusions:

  • The novel neural architecture empowered with the Hungarian algorithm offers a superior approach to paraphrase identification.
  • Effectively modeling aligned unmatched parts is key to advancing NLP tasks like paraphrase detection.
  • This research provides a promising direction for future work in sequence alignment and NLP.