Synapse
⌘+K
Synapse
PulseExploreClubsResearchersJournals
Instagram
HomeClubsExplore
June 6, 2026Journal of the Korea Society of Computer and Information

Optimizing Emotion Recognition in Korean Counseling Texts Using Parameter Efficient Fine Tuning

View Full Paper
Ask AI
Bookmark
Share

Authors

MLMyung-Suk Lee

Discussion

Loading...

Member takes

Overview

Randomized trial demonstrates improved emotion recognition in Korean counseling texts, suggesting better mental health support efficiency.

Key Points

  • This study aims to propose a parameter-efficient emotion recognition model optimized for Korean counseling dialogues.
  • Applied LoRA and QLoRA techniques on KcBERT
  • Utilized AI Hub corpus for fine-tuning
  • Compared performance with KoGPT2 using accuracy and F1-score metrics.
  • Achieved accuracy of 0.824 and F1-score of 0.812 with only 0.3% of parameters updated
  • Confirmed stable emotion expression learning through t-SNE analysis
  • Demonstrated practical deployment potential of high-performance models in resource-limited environments.

Cite This Study

Myung-Suk Lee (2026) studied this question.

synapsesocial.com/papers/6a23b9ca71a5da9775e7594fhttps://doi.org/10.9708/jksci.2026.31.05.075
View Full Paper
Ask AI
Bookmark
Share

Also Consider

Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context:

  1. 1<b>Empathy-Enhanced Small Language Models for Digital Mental Health Counseling</b>2025 · 2 citations
  2. 2A Proposed Solution for Improving Emotion Classification Accuracy2025
  3. 3Fine-grained emotion recognition using RoBERTa with low-rank adaptation and retrieval-augmented context contrastive definition pretraining masked language modeling and multi-dataset transfer learning2026 · 1 citations
  4. 4A Hybrid Random Forest+LSTM Framework for Enhanced EEG-Based Emotion Recognition2026
  5. 5Pairwise Empathetic Dialogue Generation via Automatic Large Language Model Preference Data Construction2026