基于语义对称分解哈希的快速ECG信号检测

Fast ECG Signal Detection via Symmetric Semantics Decomposition Hashing

本文提出一种语义对称分解哈希(Symmetric Semantics Decomposition Hashing,SSDH)算法来实现心电图(Electrocardiogram,ECG)信号的识别检测。SSDH首先利用ECG数据标签生成语义相似矩阵,然后对其进行离散的对称哈希分解,从而生成哈希编码库,最后利用学习的哈希函数建立海明空间与原始数据核化空间的映射关系。针对离散优化,基于语义标签,SSDH排列Hadamard矩阵,无需迭代和任何参数的调试即可快速地生成哈希编码库。在基准ECG数据集的实验结果表明SSDH可以更加快速地实现ECG信号检测,而且识别率明显优于深度模型,为心电智能终端设备提供有效的依据和决策支持。

This paper proposes a Semantic Symmetric Decomposition Hashing(SSDH)algorithm to conduct the recognition and detection of Electrocardiogram(ECG)signals.SSDH firstly employs ECG semantic label information to generate a semantic similarity matrix,then performs discrete symmetric hashing decomposition to generate a hash codebase,and finally designs a compact hashing function to establish the mapping relationship between the Hamming space and the original data kernel space.For discrete optimization,SSDH aligns the vectors of Hadamard matrix by feat of the semantic tags,and quickly generates the hash codebase without any iteration and parameter debugging.Experiments on benchmark ECG datasets show that the proposed algorithm performs favorably against the state-of-the-art deep model with negligible computation cost,which will provide effective basis and decision support for the effortless intelligent terminal equipment.