FastONN -- Python based open-source GPU implementation for Operational Neural Networks
This work provides a practical tool for researchers and practitioners working with ONNs, offering faster training and customization options, though it is incremental as it builds on existing ONN concepts.
The authors tackled the challenge of efficiently training Operational Neural Networks (ONNs) by developing FastONN, a Python-based open-source GPU library that uses a novel vectorized formulation, resulting in improved training speed and flexibility for grid-structured data.
Operational Neural Networks (ONNs) have recently been proposed as a special class of artificial neural networks for grid structured data. They enable heterogenous non-linear operations to generalize the widely adopted convolution-based neuron model. This work introduces a fast GPU-enabled library for training operational neural networks, FastONN, which is based on a novel vectorized formulation of the operational neurons. Leveraging on automatic reverse-mode differentiation for backpropagation, FastONN enables increased flexibility with the incorporation of new operator sets and customized gradient flows. Additionally, bundled auxiliary modules offer interfaces for performance tracking and checkpointing across different data partitions and customized metrics.