arXiv 2503.170612025-03-21cs.NEcs.AIcs.LG

Replay4NCL：一种面向嵌入式AI系统神经形态持续学习的高效基于记忆重放的方法

Replay4NCL: An Efficient Memory Replay-based Methodology for Neuromorphic Continual Learning in Embedded AI Systems

Electrical and Communication Engineering Department, United Arab Emirates University（阿联酋大学电子与通信工程系）；eBrain Lab, New York University（纽约大学eBrain实验室）；School of Electrical Engineering and Computer Science (SEECS), National University of Sciences and Technology（国立科学与技术大学电子工程与计算机科学学院）

Mishal Fatima Minhas, Rachmad Vidya Wicaksana Putra, Falah Awwad, Osman Hasan, Muhammad Shafique

备注：: Accepted at the 62th Design Automation Conference (DAC) 2025, June 2025, San Francisco, CA, USA
DOI：: 10.1109/DAC63849.2025.11132839

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

针对现有神经形态持续学习记忆重放方法延迟高、能耗大不适用于嵌入式AI的问题，提出Replay4NCL方法，通过压缩潜在数据并短时步重放，结合阈值与学习率调整，在SHD数据集上实现准确率提升及延迟、内存、能耗的显著优化。

摘要：

神经形态持续学习（Neuromorphic Continual Learning，NCL）范式利用脉冲神经网络（Spiking Neural Networks，SNNs）为AI系统赋予持续学习（continual learning，CL）能力，以适应动态变化的环境。当前，最先进的方法采用基于记忆重放的技术来保留旧知识，但该技术依赖长时步和压缩-解压步骤，会产生显著的延迟与能耗开销，不适用于资源严格受限的嵌入式AI系统（如移动智能体/机器人）。为解决这一问题，我们提出Replay4NCL——一种新颖的高效基于记忆重放的方法，用于在嵌入式AI系统中实现NCL。具体而言，Replay4NCL对潜在数据（即旧知识）进行压缩，随后在NCL训练阶段以短时步重放这些数据，以最小化处理延迟与能耗。为弥补脉冲减少带来的信息损失，我们调整了神经元阈值电位和学习率设置。在Spiking Heidelberg Digits（SHD）数据集的类别增量场景下的实验结果表明，Replay4NCL可保留旧知识，Top-1准确率达90.43%，优于现有最先进方法的86.22%，同时能有效学习新任务，实现4.88倍的延迟加速、20%的潜在内存节省以及36.43%的能耗节省。这些结果凸显了Replay4NCL方法进一步提升嵌入式AI系统NCL能力的潜力。

英文摘要：

Neuromorphic Continual Learning (NCL) paradigm leverages Spiking Neural Networks (SNNs) to enable continual learning (CL) capabilities for AI systems to adapt to dynamically changing environments. Currently, the state-of-the-art employ a memory replay-based method to maintain the old knowledge. However, this technique relies on long timesteps and compression-decompression steps, thereby incurring significant latency and energy overheads, which are not suitable for tightly-constrained embedded AI systems (e.g., mobile agents/robotics). To address this, we propose Replay4NCL, a novel efficient memory replay-based methodology for enabling NCL in embedded AI systems. Specifically, Replay4NCL compresses the latent data (old knowledge), then replays them during the NCL training phase with small timesteps, to minimize the processing latency and energy consumption. To compensate the information loss from reduced spikes, we adjust the neuron threshold potential and learning rate settings. Experimental results on the class-incremental scenario with the Spiking Heidelberg Digits (SHD) dataset show that Replay4NCL can preserve old knowledge with Top-1 accuracy of 90.43% compared to 86.22% from the state-of-the-art, while effectively learning new tasks, achieving 4.88x latency speed-up, 20% latent memory saving, and 36.43% energy saving. These results highlight the potential of our Replay4NCL methodology to further advances NCL capabilities for embedded AI systems.

机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。