Mobile QR Code
Title Early Classification of Time Series using SPRT and Normalizing Flow
Authors 조준희(Jun Hee Jo) ; 최계원(Kae Won Choi)
DOI https://doi.org/10.5573/ieie.2024.61.12.66
Page pp.66-69
ISSN 2287-5026
Keywords SPRT; Normalizing flow
Abstract Flow-based models learn the complex distribution of real-world data through a series of invertible transformations starting from a simple probability distribution in the latent space. In this paper, we propose a method that utilizes the conditional probability density of data with complex probability distributions using a flow-based model and applies it to the sequential probability ratio test (SPRT). The proposed method presents a new possibility for applying flow-based models to SPRT for early classification of time series data.