WaveTF: A Fast 2D Wavelet Transform for Machine Learning in Keras

The wavelet transform is a powerful tool for performing multiscale analysis and it key subroutine in countless applications, from image processing to astronomy. Recently, has extended its range of users include the ever growing machine learning community. For library be efficiently adopted this context, needs provide transformations which can integrated seamlessly already existing workflows neural networks, being able leverage same libraries run on hardware (e.g., CPU vs GPU) as rest pipeline, without impacting training evaluation performance. In paper we present WaveTF, available Keras layer, leverages TensorFlow exploit GPU parallelism used enrich workflows. To demonstrate efficiency compare raw performance against other alternative finally measure overhead causes process when an Convolutional Neural Network.