SpikingJelly: An open-source machine learning infrastructure platform for spike-based intelligence

This paper investigates the vulnerability of spiking neural networks (SNNs) and federated learning (FL) to backdoor attacks using neuromorphic data. Despite the efficiency of SNNs and the privacy advantages of FL, particularly in low-powered devices, we demonstrate that these systems are susceptible to such attacks. We first assess the viability of using FL with SNNs using neuromorphic data, showing its potential usage. Then, we evaluate the transferability of known FL attack methods to SNNs, finding that these lead to suboptimal attack performance. Therefore, we explore backdoor attacks involving single and multiple attackers to improve the attack performance. Our primary contribution is developing a novel attack strategy tailored to SNNs and FL, which distributes the backdoor trigger temporally and across malicious devices, enhancing the attack's effectiveness and stealthiness. In the best case, we achieve a 100% attack success rate, 0.13 MSE, and 98.9 SSIM. Moreover, we adapt and evaluate an existing defense against backdoor attacks, revealing its inadequacy in protecting SNNs. This study underscores the need for robust security measures in deploying SNNs and FL, particularly in the context of backdoor attacks.

show abstract

“…where Ω is the number of time steps (we set it to Ω = 16 based on [11]), P is the polarity, H is the height, and W is the width.…”

Section: Methodsmentioning

confidence: 99%

Sniper Backdoor: Single Client Targeted Backdoor Attack in Federated Learning

Abad

Paguada

Ersoy

et al. 2023

2023 IEEE Conference on Secure and Trustworthy Machine Learning (SaTML)

View full text Add to dashboard Cite

show abstract

“…A variety of gradient-based SNN libraries have been open-sourced, most of which are written in Python for syntactical ease, and several of which are built on top of commonplace deep learning packages [25,[53][54][55][56][57]. Most approaches compose primitive functions together wrapped as a spiking neuron node, where gradients are analytically calculated using reverse autodifferentiation in the backend.…”

Section: Snntorchmentioning

confidence: 99%

“…Several alternative options are available for accelerating SNNs using CUDA-based libraries. SpikingJelly provides a CuPy backend [55], GeNN uses CUDA-generated code to implement an approximate form of BPTT [45,58], and lava-dl incorporates the most commonly used functions/neurons as optimized CUDA code, while other libraries mostly depend on the deep learning package's CUDA acceleration.…”

Section: Snntorchmentioning

confidence: 99%

Exploiting deep learning accelerators for neuromorphic workloads

Sun,

Titterton,

Gopiani

et al. 2024

Neuromorph. Comput. Eng.

View full text Add to dashboard Cite

Spiking neural networks (SNNs) have achieved orders of magnitude improvement in terms of energy consumption and latency when performing inference with deep learning workloads. Error backpropagation is presently regarded as the most effective method for training SNNs, but in a twist of irony, when training on modern graphics processing units (GPUs) this becomes more expensive than non-spiking networks. The emergence of Graphcore's Intelligence Processing Units (IPUs) balances the parallelized nature of deep learning workloads with the sequential, reusable, and sparsified nature of operations prevalent when training SNNs. IPUs adopt multi-instruction multi-data (MIMD) parallelism by running individual processing threads on smaller data blocks, which is a natural fit for the sequential, non-vectorized steps required to solve spiking neuron dynamical state equations. We present an IPU-optimized release of our custom SNN Python package, snnTorch, which exploits fine-grained parallelism by utilizing low-level, pre-compiled custom operations to accelerate irregular and sparse data access patterns that are characteristic of training SNN workloads. We provide a rigorous performance assessment across a suite of commonly used spiking neuron models, and propose methods to further reduce training run-time via half-precision training. By amortizing the cost of sequential processing into vectorizable population codes, we ultimately demonstrate the potential for integrating domain-specific accelerators with the next generation of neural networks.

show abstract

“…The higher the spiking frequency, the higher the probability that the input sample belongs to the corresponding category of the spiking neuron. We used the deep learning framework PyTorch and the spiking neural network framework SpikingJelly 38 to establish the RCSNN-12 model. After conducting many preliminary experiments, we optimized the parameters and found a set of parameters that performed well.…”

Section: Sensor Typementioning

confidence: 99%

An Odor Recognition Algorithm of Electronic Noses Based on Convolutional Spiking Neural Network for Spoiled Food Identification

Xiong

Chen

et al. 2021

J. Electrochem. Soc.

View full text Add to dashboard Cite

The electronic nose is an odor detection instrument utilizing the bionic olfactory theory, and usually consisting of a gas sensor array and an odor recognition algorithm. Traditional odor recognition algorithms for electronic noses suffer from cumbersome feature extraction steps and low recognition accuracy, and some new algorithms based on deep learning have the disadvantage of insufficient computational efficiency. To solve these problems, a novel odor recognition algorithm based on convolutional spiking neural network is proposed. The network model of this algorithm is composed of convolutional spiking layers, fully-connected spiking layers, and skip connections, which combines the feature extraction ability of the residual convolutional neural network and the computational efficiency of the spiking neural network. We also established a spoiled food mixed odor dataset and a rotten fruit odor dataset for evaluation of our method and several other classic algorithms. Experimental results show that our algorithm achieves average test accuracy of 84.5% and 88.6% on these two datasets, respectively, and has better recognition capability and computational efficiency than several other classic algorithms, which indicates that our algorithm can efficiently identify the spoilage odor in food.

show abstract

SpikingJelly: An open-source machine learning infrastructure platform for spike-based intelligence

Cited by 47 publications

References 155 publications

Sniper Backdoor: Single Client Targeted Backdoor Attack in Federated Learning

Sniper Backdoor: Single Client Targeted Backdoor Attack in Federated Learning

Exploiting deep learning accelerators for neuromorphic workloads

An Odor Recognition Algorithm of Electronic Noses Based on Convolutional Spiking Neural Network for Spoiled Food Identification

Contact Info

Product

Resources

About