Spline Filters For End-to-End Deep Learning

Randall Balestriero , Romain Cosentino , Hervé Glotin (1) , Richard Baraniuk (2)

1 DYNI - DYNamiques de l’Information
2 Electrical and Computer Engineering - Rice University

We propose to tackle the problem of end-to-end learning for raw waveform signals by introducing learnable continuous time-frequency atoms. The derivation of these filters is achieved by defining a functional space with a given smoothness order and boundary conditions. From this space, we derive the parametric analytical filters. Their differentiability property allows gradient-based optimization. As such, one can utilize any Deep Neural Network (DNN) with these filters. This enables us to tackle in a front-end fashion a large scale bird detection task based on the freefield1010 dataset known to contain key challenges , such as the dimensionality of the inputs data (> 100, 000) and the presence of additional noises: multiple sources and soundscapes.

Mots clés: deep learning; spline; wavelet

Domaines: Intelligence artificielle [cs.AI]; Bio-informatique [q-bio.QM]; Vision par ordinateur et reconnaissance de formes [cs.CV]; Recherche d'information [cs.IR]; Traitement du signal et de l'image [eess.SP]; Biodiversité et Ecologie

Randall Balestriero, Romain Cosentino, Hervé Glotin, Richard Baraniuk. Spline Filters For End-to-End Deep Learning. 35th International Conference on Machine Learning, Jul 2018, stockholm, Sweden. ⟨hal-01879266⟩