I am doing data analysis. I used the wavelet transform and now I am trying the Hilbert–Huang transform (HHT). In the literature, I read that Hilbert–Huang transform (HHT) is an adaptive technique. I tried to find how it is adaptive or what is the meaning of adaptive in this context. I didn't get any answer.
The Hilbert-Huang signal decomposition possesses a non-adaptive part (the Hilbert transform) and an adaptive, or data-driven part: the decomposition of a signal as a sum for Intrinsic Mode Functions (IMF), that are obtained by a deflation process, using extrema and envelopes of the signal, recursively.
Thus, IMFs are adaptive with respect to the signal: if you change the value of one sample, the IMFs can be drastically different. For some classes of processes, namely fractional Brownian motions, the Empirical mode decompositions [can act as] as data-driven wavelet-like expansions. In other words, under some conditions, an adaptive algorithm may behave like a non-adaptive one.
There is one code for Marginal Hilbert Spectrum on Matlab Central.