Fast Sparse Connectivity Network Adaption via Meta-Learning

Jin, Bo; Ke, Cheng; Qu, Yue; Zhang, Liang; Xiao, Keli; Lu, Xinjiang; Wei, Xiaopeng

doi:10.1109/icdm50108.2020.00032

Cited by 3 publications

(4 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Reweighted LASSO [22] constructed feature connected network based on weighted penalty matrix and LASSO regression. SAMCN LASSO [8] utilized gradient meta-learning method. SAMCN Structural LASSO [8] constructed the features connectivity network based on gradient meta-learning, sparse weight penalty matrix and LASSO regression.…”

Section: Research On Chronic Disease Diagnosis Algorithmsmentioning

confidence: 99%

“…SAMCN LASSO [8] utilized gradient meta-learning method. SAMCN Structural LASSO [8] constructed the features connectivity network based on gradient meta-learning, sparse weight penalty matrix and LASSO regression. However, these methods depend on complex feature processing and calculation and can only capture linear relationships, e.g., Pearson correlation, partial correlation, etc.…”

Section: Research On Chronic Disease Diagnosis Algorithmsmentioning

confidence: 99%

“…However, they mainly explore to utilize the linear relationships. For example, both Qu et al [7] and Cheng et al [8] propose to learn the partial correlation-based connectivity matrix via either shared template regularization [7] or the sparse variational regularization [8]. Even though graph neural networks (GNNs) have been widely adopted to automatically extract the high-level complex feature correlations from multivariate time series data [9][10][11][12], present frameworks are not ideal for multivariate longitudinal data-induced chronic disease diagnosis tasks.…”

mentioning

confidence: 99%

“…Worse still, current chronic diseases (e.g., PD) classification algorithms [7,8,[13][14][15] fail to quantify the model uncertainty [16,17], which is extremely important in such high-stake healthcare analysis task. For example, if some abnormal noises are added to the multivariate time series data, the model will usually give a wrong prediction with high confidence since the final predictions are obtained only based on the output probability score; thus, it inherently is an overconfident and unreliable model [10].…”

mentioning

confidence: 99%

See 3 more Smart Citations

Exploring unsupervised multivariate time series representation learning for chronic disease diagnosis

Zhang

Wang

Zhang

et al. 2021

Int J Data Sci Anal

Self Cite

View full text Add to dashboard Cite

The application of various sensors in hospitals has enabled the widespread utilization of multivariate time series signals for chronic disease diagnosis in the data-driven world. The key challenge is how to model the complex temporal (linear and nonlinear) correlations among multiple longitudinal variables. Due to scarcity of labels in practice, unsupervised learning methods have already become indispensable. However, state-of-the-art approaches mainly focus on the extraction of linear correlation-induced feature connectivity network, e.g., Pearson correlation, partial correlation, etc. To this end, for chronic disease (e.g., Parkinson disease) diagnosis, an unsupervised representation learning method is first designed to obtain informative correlation-aware signals from multivariate time series data. At the core is a contrastive learning framework with a graph neural network (GNN) encoder to capture the inter-correlation and intra-correlation of multiple longitudinal variables. Then, the previously learned representations are sent to a simple fully connected neural network, which can be trained using fewer labels compared with end-to-end complex supervised learning models. Further, to assist the decision-making process in the high-stake chronic disease detection task, model uncertainty quantification is enabled according to evidential theory. The experimental results on two public Parkinson's disease data sets show the expressiveness of the learned embeddings, and the final lightweight classifier achieves the best performance. IntroductionWith the development of science and technology, people's living standards have been greatly improved. However, chronic diseases such as heart disease, cancer, and diabetes are still the leading causes of death and disability in many

show abstract

Section: Research On Chronic Disease Diagnosis Algorithmsmentioning

confidence: 99%

Section: Research On Chronic Disease Diagnosis Algorithmsmentioning

confidence: 99%