Building Function Approximators on top of Haar Scattering Networks
This work addresses the need for efficient and interpretable function approximators in fields like signal processing and system identification, though it appears incremental as it builds on existing Haar Scattering Networks.
The authors tackled the problem of creating general-purpose function approximators by proposing an architecture based on Haar Scattering Networks, which they demonstrated to have strong approximation and feature extraction capabilities across various domains such as signal processing and econometrics.
In this article we propose building general-purpose function approximators on top of Haar Scattering Networks. We advocate that this architecture enables a better comprehension of feature extraction, in addition to its implementation simplicity and low computational costs. We show its approximation and feature extraction capabilities in a wide range of different problems, which can be applied on several phenomena in signal processing, system identification, econometrics and other potential fields.