AI Article Synopsis

  • Spiking neural networks (SNNs) are gaining popularity for their energy efficiency and biological realism, but they still rely on high-precision values for key components like membrane potential, leading to inefficiencies in resource-limited settings.* -
  • Existing approaches to reduce these issues have resulted in significant accuracy drops, particularly when implementing low-precision representation or time step reduction.* -
  • The proposed SpQuant-SNN addresses these challenges by integrating an integer-only quantization scheme, spatial-channel pruning, and a self-adaptive learnable threshold to improve performance while maintaining low computational demand and memory usage.*

Article Abstract

Spiking neural networks (SNNs) have received increasing attention due to their high biological plausibility and energy efficiency. The binary spike-based information propagation enables efficient sparse computation in event-based and static computer vision applications. However, the weight precision and especially the membrane potential precision remain as high-precision values (e.g., 32 bits) in state-of-the-art SNN algorithms. Each neuron in an SNN stores the membrane potential over time and typically updates its value in every time step. Such frequent read/write operations of high-precision membrane potential incur storage and memory access overhead in SNNs, which undermines the SNNs' compatibility with resource-constrained hardware. To resolve this inefficiency, prior works have explored the time step reduction and low-precision representation of membrane potential at a limited scale and reported significant accuracy drops. Furthermore, while recent advances in on-device AI present pruning and quantization optimization with different architectures and datasets, simultaneous pruning with quantization is highly under-explored in SNNs. In this work, we present , a fully-quantized spiking neural network with , enabling the end-to-end low precision with significantly reduced operations on SNN. First, we propose an integer-only quantization scheme for the membrane potential with a stacked surrogate gradient function, a simple-yet-effective method that enables the smooth learning process of quantized SNN training. Second, we implement spatial-channel pruning with membrane potential prior, toward reducing the layer-wise computational complexity, and floating-point operations (FLOPs) in SNNs. Finally, to further improve the accuracy of low-precision and sparse SNN, we propose a self-adaptive learnable potential threshold for SNN training. Equipped with high biological adaptiveness, minimal computations, and memory utilization, SpQuant-SNN achieves state-of-the-art performance across multiple SNN models for both event-based and static image datasets, including both image classification and object detection tasks. The proposed SpQuant-SNN achieved up to 13× memory reduction and >4.7× FLOPs reduction with < 1.8% accuracy degradation for both classification and object detection tasks, compared to the SOTA baseline.

Download full-text PDF

Source
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC11408473PMC
http://dx.doi.org/10.3389/fnins.2024.1440000DOI Listing

Publication Analysis

Top Keywords

membrane potential
28
spiking neural
12
potential
9
precision membrane
8
neural networks
8
high biological
8
event-based static
8
time step
8
pruning quantization
8
snn propose
8

Similar Publications

Mitochondria-targeted antioxidant MitoQ improves the quality of low temperature-preserved yak semen via alleviating oxidative stress.

Anim Reprod Sci

December 2024

Key Laboratory for Animal Science of National Ethnic Affairs Commission, Southwest Minzu University, Chengdu 610041, China; Key Laboratory of Qinghai-Tibetan Plateau Animal Genetic Resource Reservation and Exploitation of Ministry of Education, Southwest Minzu University, Chengdu 610041, China. Electronic address:

Low-temperature preservation of yak semen during transportation and conservation is crucial to accelerate yak breeding. The effects of low-temperature cooling on yak semen quality, however, are poorly understood. This study aimed to determine the dose-dependent effect of mitochondria-targeted antioxidant "MitoQ" on the motility, oxidative status, and mitochondrial function of yak semen during low-temperature preservation.

View Article and Find Full Text PDF

The integration of membrane separation with heterogeneous advanced oxidation processes is a prospective strategy for the elimination of contaminants during wastewater treatment. Fe-based catalysts and the green oxidant peracetic acid (PAA) are desirable candidates for the development of catalytic membranes because they are environmentally friendly. However, the construction of catalytic ceramic membranes (CMs) modified with efficient Fe-based catalysts that generate increased amounts of high-valent Fe-O species during PAA activation for the degradation of specific pollutants, especially during instantaneous membrane filtration, remains challenging.

View Article and Find Full Text PDF

Background And Objective: Vancomycin is a glycopeptide antibiotic used for the treatment of severe gram-positive infections. Despite decades of clinical experience, optimized dosing for vancomycin in pediatric populations still warrants further investigation. Patients admitted to the pediatric intensive care unit (PICU) after cardiac surgery are often treated with vancomycin in case of (suspected) infection.

View Article and Find Full Text PDF

cGAS-STING pathway stands at the forefront of innate immunity and plays a critical role in regulating adaptive immune responses, making it as a key orchestrator of anti-tumor immunity. Despite the great potential, clinical outcomes with cGAS-STING activators have been disappointing due to their unfavorable in vivo fate, signaling an urgent need for innovative solutions to bridge the gap in clinical translation. Recent advancements in nanotechnology have propelled cGAS-STING-targeting nanomedicines to the cutting-edge of cancer therapy, leveraging precise drug delivery systems and multifunctional platforms to achieve remarkable region-specific biodistribution and potent therapeutic efficacy.

View Article and Find Full Text PDF

Objective: The aim of this study was to investigate the dynamic changes in QTc interval duration among patients with COVID-19 infection before, during, and after infection, in order to assess the short- and potential long-term impact of COVID-19 on cardiac electrophysiology.

Methods: A retrospective analysis was conducted on 303 inpatients diagnosed with COVID-19 who visited a tertiary Grade A hospital in China between August 2022 and December 2023. Inclusion criteria required patients to have at least two electrocardiogram (ECG) recordings at three specific time points: before COVID-19 infection, during acute infection, and after recovery (more than one month post-infection).

View Article and Find Full Text PDF

Want AI Summaries of new PubMed Abstracts delivered to your In-box?

Enter search terms and have AI summaries delivered each week - change queries or unsubscribe any time!