Incremental learning for audio classification with Hebbian Deep Neural Networks
For audio classification, this work introduces a biologically inspired incremental learning approach that improves stability and accuracy, though it is incremental in nature.
The authors applied Hebbian learning to sound classification and proposed a kernel plasticity method for incremental learning, achieving 76.3% accuracy on ESC-50 over five steps, outperforming a baseline without plasticity (68.7%).
The ability of humans for lifelong learning is an inspiration for deep learning methods and in particular for continual learning. In this work, we apply Hebbian learning, a biologically inspired learning process, to sound classification. We propose a kernel plasticity approach that selectively modulates network kernels during incremental learning, acting on selected kernels to learn new information and on others to retain previous knowledge. Using the ESC-50 dataset, the proposed method achieves 76.3% overall accuracy over five incremental steps, outperforming a baseline without kernel plasticity (68.7%) and demonstrating significantly greater stability across tasks.