Skip to content

gcliu1982/CNNM

Repository files navigation

Convolution Nuclear Norm for Data Forecasting and Completion

matlab code for tensor-valued time series forecasting and completion. The code includes two algorithms, DFT_L1 (Discrete Fourier Transform based L1 minimization) and CNNM (Convolution Nuclear Norm Minimization).

Please run the following three routines to get started:

demo_1Dtimeseries.m ---- a demo for 1D time serires forecasting

demo_imagecompletion.m ------- a demo for image completion

demo_videoprediction.m ------- a demo for video prediction, i.e., predicting the next q image frames of a given sequence of p frames.

Reference:

[1] Guangcan Liu, Wayne Zhang, Recovery of Future Data via Convolution Nuclear Norm Minimization, Arxiv, 2019

About

matlab code for tensor-valued time series forecasting and completion

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages