![]() Learning semantic-rich representations from raw unlabeled time series data is critical for downstream tasks such as classification and forecasting. ![]() Contrastive learning has recently shown its promising representation learning capability in the absence of expert annotations. However, existing contrastive approaches generally treat each instance independently, which leads to false negative pairs that share the same semantics. To tackle this problem, we propose MHCCL, a Masked Hierarchical Cluster-wise Contrastive Learning model, which exploits semantic information obtained from the hierarchical structure consisting of multiple latent partitions for multivariate time series. Motivated by the observation that fine-grained clustering preserves higher purity while coarse-grained one reflects higher-level semantics, we propose a novel downward masking strategy to filter out fake negatives and supplement positives by incorporating the multi-granularity information from the clustering hierarchy. In addition, a novel upward masking strategy is designed in MHCCL to remove outliers of clusters at each partition to refine prototypes, which helps speed up the hierarchical clustering process and improves the clustering quality. We conduct experimental evaluations on seven widely-used multivariate time series datasets. The results demonstrate the superiority of MHCCL over the state-of-the-art approaches for unsupervised time series representation learning.ĭeep neural networks currently achieve state-of-the-art performance in many multivariate time series classification (MTSC) tasks, which are crucial for various real-world applications. However, the black-box characteristic of deep learning models impedes humans from obtaining insights into the internal regulation and decisions made by classifiers. To the north and the Atlantic Ocean to the south, Augusta is a thriving communityīuilt on a solid foundation of local pride and artistic eccentricity.Existing explainability research generally requires constructing separate explanation models to work with deep learning models or process their results, thus calling for additional development efforts. Serving as a halfway point between the Appalachian Mountains Live music and performances, unique food and cultural offerings, top-rated schoolsĪugusta University’s primary campuses in Augusta are situated on the southern banks Retaining access to the rich resources of Georgia’s second-largest city, including Additionally, opportunities to impart knowledgeĪnd application abound in the Hull College of Business Katherine Reese Pamplin College of Arts, Humanities, and Social Sciences the Colleges of Education, Science and Mathematics, Allied Health Sciences, and Nursing and The Graduate School.Īt Augusta, our faculty and staff have the unique opportunity to be part of an intimateĬampus experience with small-class sizes and a diverse community of learners, while Medical College of Georgia, the state’s only dental school, The Dental College of Georgia, and the state-of-the art School of Computer and Cyber Sciences located in the Georgia Cyber Center. Our colleges and schools include the nation’s eighth-largest medical school, the Practice, and provide support to our students and patients across 8 colleges and 2 schools, at our 2 libraries, and in our world class medical center. At Augusta University, faculty and staff work together to teach, conduct research,
0 Comments
Leave a Reply. |