Machine Learning

Basic Filters for Convolutional Neural Networks: Training or Design?

This topic contains 0 replies, has 1 voice, and was last updated by  arXiv 2 months, 2 weeks ago.


  • arXiv
    5 pts

    Basic Filters for Convolutional Neural Networks: Training or Design?

    When convolutional neural networks are used to tackle learning problems based on time series, e.g., audio data, raw one-dimensional data are commonly pre-processed to obtain spectrogram or mel-spectrogram coefficients, which are then used as input to the actual neural network. In this contribution, we investigate, both theoretically and experimentally, the influence of this pre-processing step on the network’s performance and pose the question, whether replacing it by applying adaptive or learned filters directly to the raw data, can improve learning success. The theoretical results show that approximately reproducing mel-spectrogram coefficients by applying adaptive filters and subsequent time-averaging is in principle possible. On the other hand, extensive experimental work leads to the conclusion, that the invariance induced by mel-spectrogram coefficients is both desirable and hard to infer by the learning process. Thus, the results achieved by adaptive end-to-end learning approaches are close to but slightly worse than results achieved by state-of-the-art reference architectures using standard input coefficients derived from the spectrogram.

    Basic Filters for Convolutional Neural Networks: Training or Design?
    by Monika Doerfler, Thomas Grill, Roswitha Bammer, Arthur Flexer
    https://arxiv.org/pdf/1709.02291v1.pdf

You must be logged in to reply to this topic.