Features Fusion Framework for Multimodal Irregular Time-series Events

被引:1
|
作者
Tang, Peiwang [1 ,2 ]
Zhang, Xianchao [3 ,4 ]
机构
[1] Univ Sci & Technol China, Inst Adv Technol, Hefei 230026, Peoples R China
[2] Jiaxing Univ, G60 STI Valley Ind Innovat Inst, Jiaxing 314001, Peoples R China
[3] Jiaxing Univ, Key Lab Med Elect & Digital Hlth Zhejiang Prov, Jiaxing 314001, Peoples R China
[4] Jiaxing Univ, Engn Res Ctr Intelligent Human Hlth Situat Awaren, Jiaxing 314001, Peoples R China
关键词
Features fusion; LSTM; Multimodal; Time-series;
D O I
10.1007/978-3-031-20862-1_27
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Some data from multiple sources can be modeled as multimodal time-series events which have different sampling frequencies, data compositions, temporal relations and characteristics. Different types of events have complex nonlinear relationships, and the time of each event is irregular. Neither the classical Recurrent Neural Network (RNN) model nor the current state-of-the-art Transformer model can deal with these features well. In this paper, a features fusion framework for multimodal irregular time-series events is proposed based on the Long Short-Term Memory networks (LSTM). Firstly, the complex features are extracted according to the irregular patterns of different events. Secondly, the nonlinear correlation and complex temporal dependencies relationship between complex features are captured and fused into a tensor. Finally, a feature gate are used to control the access frequency of different tensors. Extensive experiments on MIMIC-III dataset demonstrate that the proposed framework significantly outperforms to the existing methods in terms of AUC (the area under Receiver Operating Characteristic curve) and AP (Average Precision).
引用
收藏
页码:366 / 379
页数:14
相关论文
共 50 条
  • [41] Irregular Trend Finder: Visualization tool for analyzing time-series big data
    Takeda, Shinnosuke
    Kobayashi, Aimi
    Kobayashi, Hiroaki
    Okubo, Saori
    Misue, Kazuo
    2012 IEEE CONFERENCE ON VISUAL ANALYTICS SCIENCE AND TECHNOLOGY (VAST), 2012, : 305 - 306
  • [42] Multimodal Time Series Data Fusion Based on SSAE and LSTM
    Zhu, Qiding
    Zhang, Shukui
    Zhang, Yang
    Yu, Chunqing
    Dang, Mengli
    Zhang, Li
    2021 IEEE WIRELESS COMMUNICATIONS AND NETWORKING CONFERENCE (WCNC), 2021,
  • [43] Fuzzy time-series prediction model based on text features and network features
    Liu, Zeguang
    Li, Yao
    Liu, Huilin
    NEURAL COMPUTING & APPLICATIONS, 2023, 35 (05): : 3639 - 3649
  • [44] Fuzzy time-series prediction model based on text features and network features
    Zeguang Liu
    Yao Li
    Huilin Liu
    Neural Computing and Applications, 2023, 35 : 3639 - 3649
  • [45] Time-series representation learning via Time-Frequency Fusion Contrasting
    Zhao, Wenbo
    Fan, Ling
    FRONTIERS IN ARTIFICIAL INTELLIGENCE, 2024, 7
  • [46] Time-series Classification Using Neural Bag-of-Features
    Passalis, Nikolaos
    Tsantekidis, Avraam
    Tefas, Anastasios
    Kanniainen, Juho
    Gabbouj, Moncef
    Iosifidis, Alexandros
    2017 25TH EUROPEAN SIGNAL PROCESSING CONFERENCE (EUSIPCO), 2017, : 301 - 305
  • [47] Utilizing Expert Features for Contrastive Learning of Time-Series Representations
    Nonnenmacher, Manuel
    Oldenburg, Lukas
    Steinwart, Ingo
    Reeb, David
    INTERNATIONAL CONFERENCE ON MACHINE LEARNING, VOL 162, 2022,
  • [48] SOME LONG RUN FEATURES OF DYNAMIC TIME-SERIES MODELS
    CURRIE, D
    ECONOMIC JOURNAL, 1981, 91 (363): : 704 - 715
  • [49] A Framework for Periodic Outlier Pattern Detection in Time-Series Sequences
    Rasheed, Faraz
    Alhajj, Reda
    IEEE TRANSACTIONS ON CYBERNETICS, 2014, 44 (05) : 569 - 582
  • [50] DERIVATIVE-AUGMENTED FEATURES AS A DYNAMIC MODEL FOR TIME-SERIES
    Baggenstoss, Paul M.
    2015 23RD EUROPEAN SIGNAL PROCESSING CONFERENCE (EUSIPCO), 2015, : 958 - 962