Including STDP to eligibility propagation in multi-layer recurrent spiking neural networks
This work addresses the problem of efficient training for low-power neuromorphic hardware, but it is incremental as it builds on existing e-prop methods with minor modifications.
The study tackled the lack of competitive learning algorithms for training spiking neural networks (SNNs) by analyzing eligibility propagation (e-prop) with STDP-like behavior, finding that including STDP improved classification performance for the ALIF neuron model but not for the Izhikevich neuron, and that single-layer recurrent SNNs outperformed multi-layer variants.
Spiking neural networks (SNNs) in neuromorphic systems are more energy efficient compared to deep learning-based methods, but there is no clear competitive learning algorithm for training such SNNs. Eligibility propagation (e-prop) offers an efficient and biologically plausible way to train competitive recurrent SNNs in low-power neuromorphic hardware. In this report, previous performance of e-prop on a speech classification task is reproduced, and the effects of including STDP-like behavior are analyzed. Including STDP to the ALIF neuron model improves the classification performance, but this is not the case for the Izhikevich e-prop neuron. Finally, it was found that e-prop implemented in a single-layer recurrent SNN consistently outperforms a multi-layer variant.