KPIs missing value interpolation method based on improved Transform and masking training

The invention provides a KPIs missing value interpolation method based on improved Transform and masking training. The KPIs missing value interpolation method comprises the following steps: constructing a KPIs time sequence; carrying out position coding on the time sequence of the KPIs through Posit...

Full description

Saved in:
Bibliographic Details
Main Authors WANG PEIPENG, ZHAO YUN, ZHANG XIUGUO, WANG TENGLONG, CAO ZHIYING, SHANG ZIJING
Format Patent
LanguageChinese
English
Published 31.01.2023
Subjects
Online AccessGet full text

Cover

Loading…
More Information
Summary:The invention provides a KPIs missing value interpolation method based on improved Transform and masking training. The KPIs missing value interpolation method comprises the following steps: constructing a KPIs time sequence; carrying out position coding on the time sequence of the KPIs through Position Encoding, and carrying out position coding on the time sequence of the KPIs through Position Encoding; performing para-position addition on the KPIs time sequence and the data set after position coding; a sequence obtained through para-position addition is input into an interpolation model based on an improved Transform, any stacked layer comprises a multi-head attention network and a feedforward network, and diagonal masks are applied in the multi-head attention network; and obtaining an output sequence of an interpolation model based on an improved Transform, reducing the dimension of the output sequence to be the same as the dimension of the KPIs time sequence through Linear linear transformation to obtain a
Bibliography:Application Number: CN202211379295