Multimodal Language Processing Lab.
at Seoul National University of Science and Technology
서울과학기술대학교 멀티모달 자연어처리 연구실 (임경태 교수와 MLP랩)
mlp.seoultech.ac.kr
News
(2024, Mar.) Our paper X-LLaVA: Optimizing Bilingual Large Vision-Language Alignment is accepted to Findings of NAACL 2024 (BK 우수학술대회)
(2024, Feb.) Our paper Optimizing Language Augmentation for Multilingual Large Language Models: A Case Study on Korean is accepted to LREC-COLING 2024 (BK 우수학술대회)
(2024, Feb.) Our paper Towards Standardized Annotation and Parsing for Korean FrameNet is accepted to LREC-COLING 2024 (BK 우수학술대회)
(2024, Feb.) Our paper A Linguistically-Informed Annotation Strategy for Korean Semantic Role Labeling is accepted to LREC-COLING 2024 (BK 우수학술대회)
(2024, Jan.) Our paper "Sow Posture Analysis and Estrus Prediction using Closed-circuit Television Cameras" has been accepted by IEEE Access (SCIE).
(2024, Jan.) Our paper "A Novel Communication Framework of SMRs: A Prototype Development of a NLP-based System" has been accepted by Nuclear Technology (SCIE).
(2023, Dec.) Our paper "BOK-VQA: Bilingual Outside Knowledge-based Visual Question Answering via Graph Representation Pretraining" is accepted to AAAI2024 (BK우수) [Link]
(2023, Nov.) Our lab won 국회도서관관장상 at "The Competition of Data*AI Analysis in 2023". Congrats!
(2023, Oct. Our paper "Lightweight Image Steganalysis with Block-wise Pruning" is accepted to Scientific Report (SCIE).
(2023, Aug.) Our paper "Data Augmentation of X-ray Images for Automatic Cargo Inspection of Nuclear Items" is accepted to Sensors (SCIE).
(2023, July) Our paper "Visual Question Answering Empowered by Knowledge Graph Embedding" has received the Best Presentation Paper Award at the KCC2023.
(2023, July) Our paper "Korean Named Entity Recognition Based on Language Specific Features" is accepted to Natural Language Engineering (A&HCI Q1) [Link].
(2023, June) Five papers are accepted to KCC2023 with the topics VQA, Document Classification, Visual Reasoning, and Steganography with 김상민, 김민준, 송승우, 육정훈, 임현석, 최창수
(2023, June) Our paper "Teddysum at MEDIQA-Chat 2023: an analysis of fine-tuning strategy for long dialog summarization" is accepted to at ACL 2023 workshop
(2023, May) Our paper "K-UniMorph: Korean Universal Morphology and its Feature Schema" is accepted to ACL 2023 (BK 우수학술대회)
(2023, Feb.) Our paper "Semantic Similarity-based Visual Reasoning without Language Information" is accepted to ICAIIC 2023
(2022. Aug.) Our paper "Yet Another Format of Universal Dependencies for Korean" is accepted to COLING 2022 (BK 우수학술대회) [Link]
(2022. Aug.) Our paper "Efficient Multilingual Multi-modal Pre-training through Triplet Contrastive Loss" is accepted to COLING 2022 (BK 우수학술대회) [Link]
(2022. June) MinJun Kim received Presidential Science Scholarship(대통령 과학 장학금) [Link]
(2022. June) Our paper "Neural automated writing evaluation for Korean L2 writing" is accepted to Natural Language Engineering [Link]
(2022. Mar) Our paper "Real-world Sentence Boundary Detection using Multi-Task Learning: A case study on French" is accepted to Natural Language Engineering [Link]
Introduction
What we do: AAI Lab pursues research on Language Resources and Applied Artificial Intelligence (AAI) techniques.
What we research: We are mainly studying Natural Language Processing (NLP) and Multimodal Systems.
Research Topics: Large Language Model (LLM), Vision-Language Model (VLM), Machine Translation, and Dependency Parsing.
We're looking for colleagues (PhD, M.S., Undergraduate) to study artificial intelligence technology that's more realistic and fun.
챗봇, 음성인식, 번역기와 같은 실용적인 응용인공지능을 함께 연구할 학,석,박 학생을 모집하고 있습니다.
Research
서울과학기술대학교 멀티모달 언어처리 연구실(MLP Lab)은 (1) 한국어 언어자원 설계 (2) 전통 자연어처리 (3) 멀티모달 초거대 언어모델을 연구합니다
Large Language Models
Korean Language Resource Evaluation (효과적인 한국어 언어자원 설계 및 구축 방법연구)
Language Model for Dialogue System (Chatbot)
Vision-Language Model
Vision-Langauge-Action Model
Traditional NLP systems. Morphological Analysis (형태소분석), Dependency Parsing (의존구조분석)
Large Language Model (LLM) & dataset based on Linguistic features
Lead of the Bllossom project. (https://www.bllossom.ai) (https://huggingface.co/MLP-KTLim/llama-3-Korean-Bllossom-8B)
Lead of the Universal Korean Language Resource project. (https://sites.google.com/view/universal-korean)
Visual language model & Video Story Understanding
Visual Question Answering (VQA) is a question answering system that answers user's questions based on the input image and question.
Bllossom-V: Vision-Language Model
What we have done (KBVQA Chatbot)
Applied Artificial Intelligence on Robotics
Vision-Language-Action Model
Control of Robots Based on Speech Teaching Command
Nvidia Jetson Nano, Jet Racer, Jet racer pro
Spot (Boston Dynamics)
