We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs

Change to browse by:

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Neural and Evolutionary Computing

Title: Training Energy-Efficient Deep Spiking Neural Networks with Single-Spike Hybrid Input Encoding

Abstract: Spiking Neural Networks (SNNs) have emerged as an attractive alternative to traditional deep learning frameworks, since they provide higher computational efficiency in event driven neuromorphic hardware. However, the state-of-the-art (SOTA) SNNs suffer from high inference latency, resulting from inefficient input encoding and training techniques. The most widely used input coding schemes, such as Poisson based rate-coding, do not leverage the temporal learning capabilities of SNNs. This paper presents a training framework for low-latency energy-efficient SNNs that uses a hybrid encoding scheme at the input layer in which the analog pixel values of an image are directly applied during the first timestep and a novel variant of spike temporal coding is used during subsequent timesteps. In particular, neurons in every hidden layer are restricted to fire at most once per image which increases activation sparsity. To train these hybrid-encoded SNNs, we propose a variant of the gradient descent based spike timing dependent back propagation (STDB) mechanism using a novel cross entropy loss function based on both the output neurons' spike time and membrane potential. The resulting SNNs have reduced latency and high activation sparsity, yielding significant improvements in computational efficiency. In particular, we evaluate our proposed training scheme on image classification tasks from CIFAR-10 and CIFAR-100 datasets on several VGG architectures. We achieve top-1 accuracy of $66.46$\% with $5$ timesteps on the CIFAR-100 dataset with ${\sim}125\times$ less compute energy than an equivalent standard ANN. Additionally, our proposed SNN performs $5$-$300\times$ faster inference compared to other state-of-the-art rate or temporally coded SNN models.
Comments: Extended version of arXiv:2107.11979
Subjects: Neural and Evolutionary Computing (cs.NE)
Cite as: arXiv:2107.12374 [cs.NE]
  (or arXiv:2107.12374v1 [cs.NE] for this version)

Submission history

From: Gourav Datta [view email]
[v1] Mon, 26 Jul 2021 06:16:40 GMT (6063kb,D)

Link back to: arXiv, form interface, contact.