De-shape Short-Time Fourier Transform
ds-STFT (or ds-SST) is a nonlinear type time-frequency analysis that combines cepstrum into the time-frequency analysis (See Lin & Su & Wu, JFAA, 2017). This is the code that is used to generate the figures in the JFAA 2017 paper. If you found any bug, please let me know.



Blaschke decomposition
Blaschke decomposition (BKD) (or called adaptive Fourier transform in some literature) is a data-driven signal decomposition algorithm, which is based on the complicated nonlinear behavior of the zeros. You can download the code we used to generate the BKD paper here (updated July 08 2016):



ConceFT
ConceFT is a new data-adaptive time-frequency analysis tool, which could be applied to different time-frequency representations, like SST and other reassignment techniques. You can download the code we used to generate the conceFT paper here:



Tycoon
Tycoon is a new data-adaptive time-frequency analysis tool based on the convex optimization approach. You can download the code we used to generate the Tycoon paper here:



Synchrosqueezing Transform (SST)

SST is a data-adaptive time-frequency analysis tool aiming to deal with dynamical oscillatory signals. It has been proved to be robust to several different kinds of noise (ranging from white noise, ARMA to GARCH), able to be implemented in real-time and applied to several different kinds of practical datasets.  

[official code] You can find the latest official code in Eugene’s homepageonly the CWT version is available now (Matlab).  This code is no longer maintained by anyone!

You can download the Matlab code I used to analyze data (CWT and STFT based SST. Here is the link https://drive.google.com/file/d/0B32sHHxGemopdGV2MjVpWWRDRFU/view?usp=sharing

The code attached below allows you to generate the figures and tables for the paper ``Nonparametric  and adaptive modeling of dynamic seasonality and trend with heteroscedastic and dependent errors''. 


Blending operator 

Blending operator is a real-time interpolation scheme which intrinsically is a ``local spline interpolation'' algorithm. The purpose is to analyze a non-uniform sampled signal in the real-time manner. The code can be downloaded here:

https://drive.google.com/file/d/0B32sHHxGemopTUQ3dVdUN1dSamM/view?usp=sharing



Vector Diffusion Map (VDM)

VDM is a data-adaptive high dimensional data analysis tool aiming to capture the geometric structure of a given high dimensional point clouds. In particular, we take the frame bundle structure of the underlying manifold structure into account. 

Matlab code (my private unofficial code) can be downloaded below. Notice that you need tstool to run the nearest neighbors search. 


Manifold Adaptive Local Linear Regression Estimator (MALLER)

MALLER is a non-parametric regression framework generalizing local linear regression to the manifold setup. It has been shown that MALLER shares the same benefits as that of local linear regression: it is design-adaptive, its boundary behavior is good, it has the optimal convergence rate, and it allows us to estimate the gradient of the response function. It depends on the tangent bundle structure of the underlying manifold, and is potential to build up a Laplace-Beltrami operator with the non-Neuyman's boundary condition.

Matlab code (my private unofficial code) can be downloaded below, which . Notice that you need tstool to run the nearest neighbors search.

ċ
MALLERv01.zip
(29k)
Hau-tieng Wu,
Apr 25, 2013, 10:29 PM
ċ
SST_v03.zip
(371k)
Hau-tieng Wu,
Apr 28, 2013, 1:02 AM
ċ
VDM_v01.tar.gz
(7k)
Hau-tieng Wu,
Apr 28, 2013, 12:57 AM