Eunju Park

Hi! there👋🏻 I am passionate about advancing artificial intelligence to bridge the gap between modalities, enabling machines to perceive, understand, and interact with the world more intuitively.

Recently, I have been exploring Multi-modal Hallucination in Vision-Language Models, focusing on contrastive decoding improvements and enhanced visual perception to reduce uncertainty in real-world applications.

🔥 Key Areas : Multi-modal Learning LLM Computer Vision Vison-Language Models

✉️  E-mail : [email protected] 📞  Tel : +82-10-9041-2834 🔗 Site : https://pej0918.github.io/

👩‍💻 Internships

$\color{#43515c}\rule{361px}{1.5px}$

ETRI

한국전자통신연구원

Language Intelligence Lab

<aside> 💡 2024.07 ~ 2024.08 (2개월) | 동계 연구연수생

• Atomic Fact Validation을 위한 사실성 검증 작업
• 데이터셋 구축 및 검증
• 프롬프트 튜닝을 통한 성능 최적화
• 모델 사실성 평가를 위한 복합 Accuracy 지표 설정
  • Soft Acc. & Hard Acc.

***▶️ Details***

</aside>

🔎 Research Experiences

$\color{#43515c}\rule{361px}{1.5px}$

🧩Multi-modal

🖼️ + 🔊

• Robust Audio-Visual Classification End-to-End Framework under Uncertain Missing Modality using Prompt Learning

  2024.11 - 2025.01  | [code]

  ***▶️ Details***

🖼️ + 🔠

• Multi-modal Template-Based Learning for Few-Shot Visual Grounding

  2024.07 - 2024.10  | [code]

  ***▶️ Details***

• Question-Aware Prompting and Multi-layer Co-Attention for Enhanced Knowledge-Based Visual Question Answering

  2024.05 - 2024.08  | [code]

  ***▶️ Details***

⚠️ Anomaly Detection