Burstprop for Learning in Spiking Neuromorphic Hardware
Stuck, M.; Naud, R.
Show abstract
The need for energy-efficient solutions in Deep Neural Network (DNN) applications has led to a growing interest in Spiking Neural Networks (SNNs) implemented in neuromorphic hardware. The Burstprop algorithm enables online and local learning in hier-archical networks, and therefore can potentially be implemented in neuromorphic hardware. This work presents an adaptation of the algorithm for training hierarchical SNNs on MNIST. Our implementation requires an order of magnitude fewer neurons than the previous ones. While Burstprop outper-forms Spike-timing dependent plasticity (STDP), it falls short compared to training with backpropagation through time (BPTT). This work establishes a foundation for further improvements in the Burst-prop algorithm, developing such algorithms is essential for achieving energy-efficient machine learning in neuromorphic hardware.
Matching journals
The top 6 journals account for 50% of the predicted probability mass.
Similar papers in this journal
- A scalable implementation of the recursive least-squares algorithm for training spiking neural networks 94%
- Efficient parameter calibration and real-time simulation of large scale spiking neural networks with GeNN and NEST 94%
- A quick and easy way to estimate entropy and mutual information for neuroscience 92%
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.