인문학
사회과학
자연과학
공학
의약학
농수해양학
예술체육학
복합학
DBpia AI
AI 에이전트
AI 아이디어
AI 리더
AI 뷰어
Citeasy
크롬 서지관리 다운로드
개인구독
소속 기관이 없으신 경우, 개인 정기구독을 하시면 저렴하게
논문을 무제한 열람 이용할 수 있어요.
지원사업
학술연구/단체지원/교육 등 연구자 활동을 지속하도록 DBpia가 지원하고 있어요.
커뮤니티
연구자들이 자신의 연구와 전문성을 널리 알리고, 새로운 협력의 기회를 만들 수 있는 네트워킹 공간이에요.
피인용 0
초록·키워드
This paper proposes a novel approach for detecting anomalies in time series data with cyclic irregularities using a multimodal large language model. While previous methods utilizing image- based representations of time series have shown effectiveness in detecting point-wise and trend- based anomalies, they exhibit clear limitations in identifying structural deviations from periodicity, known as cyclic anomalies.
To address this limitation, we convert time series data into multimodal inputs consisting of plot images and spectrograms generated via Short-Time Fourier Transform. Additionally, we enhance the model's in- context learning capabilities by retrieving similar past time series from a database using Dynamic Time Warping and presenting them as few-shot examples.
Experimental results demonstrate that our approach significantly outperforms both zero-shot and randomly constructed few-shot baselines, as well as traditional time series anomaly detection algorithms. This study empirically shows that a multimodal large language model can serve as an effective tool for anomaly detection in time series data, particularly for detecting complex cyclic anomalies.
To address this limitation, we convert time series data into multimodal inputs consisting of plot images and spectrograms generated via Short-Time Fourier Transform. Additionally, we enhance the model's in- context learning capabilities by retrieving similar past time series from a database using Dynamic Time Warping and presenting them as few-shot examples.
Experimental results demonstrate that our approach significantly outperforms both zero-shot and randomly constructed few-shot baselines, as well as traditional time series anomaly detection algorithms. This study empirically shows that a multimodal large language model can serve as an effective tool for anomaly detection in time series data, particularly for detecting complex cyclic anomalies.
본문·목차
인공지능 문자 인식 모델을 통해 추출된 텍스트로, 일부 오타나 오류가 포함될 수 있으나 지속적으로 개선 중입니다.
오류를 발견하셨다면 해당 부분을 드래그한 후 ' 를 통해 신고해주세요.
오류를 발견하셨다면 해당 부분을 드래그한 후 ' 를 통해 신고해주세요.
최근 본 자료 전체보기
UCI(KEPA) : I410-151-25-02-093763242
