In this paper, we propose a phase shift deep neural network (PhaseDNN) which provides a wideband convergence in approximating a high dimensional function during its training of the network. The PhaseDNN utilizes the fact that many DNN achieves convergence in the low frequency range first, thus, a series of moderately-sized of DNNs are constructed and trained in parallel for ranges of higher frequencies. With the help of phase shifts in the frequency domain, implemented through a simple phase factor multiplication on the training data, each DNN in the series will be trained to approximate the target function's higher frequency content over a specific range. Due to the phase shift, each DNN achieves the speed of convergence as in the low frequency range. As a result, the proposed PhaseDNN system is able to convert wideband frequency learning to low frequency learning, thus allowing a uniform learning to wideband high dimensional functions with frequency adaptive training. Numerical results have demonstrated the capability of PhaseDNN in learning information of a target function from low to high frequency uniformly.
[1]
Zhi-Qin John Xu,et al.
Understanding training and generalization in deep learning by Fourier analysis
,
2018,
ArXiv.
[2]
D. Brandt,et al.
Multi-level adaptive solutions to boundary-value problems math comptr
,
1977
.
[3]
Ingrid Daubechies,et al.
Ten Lectures on Wavelets
,
1992
.
[4]
Holger Wendland,et al.
Scattered Data Approximation: Conditionally positive definite functions
,
2004
.
[5]
Zheng Ma,et al.
Frequency Principle: Fourier Analysis Sheds Light on Deep Neural Networks
,
2019,
Communications in Computational Physics.
[6]
Jianzhong Wang,et al.
Adaptive multiresolution collocation methods for initial boundary value problems of nonlinear PDEs
,
1996
.