# Event-triggered Implicit Perturbation for Zeroth-Order Fine-Tuning of Spiking Transformers

arXiv: https://arxiv.org/abs/2608.21223 · HTML: https://mlsanity.com/paper/2608.21223/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 20 of its 50 references that are in the map.
- First posted on arXiv 21 Aug 2026.

## Builds on (most cited first, 20 of 20)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [A Survey on Visual Transformer](https://mlsanity.com/paper/2012.12556/) (2020)
- [Fine-Tuning Language Models with Just Forward Passes](https://mlsanity.com/paper/2305.17333/) (2023)
- [Visual Transformers: Token-based Image Representation and Processing for Computer Vision](https://mlsanity.com/paper/2006.03677/) (2020)
- [A Primer on Zeroth-Order Optimization in Signal Processing and Machine Learning](https://mlsanity.com/paper/2006.06224/) (2020)
- [Revisiting Zeroth-Order Optimization for Memory-Efficient LLM Fine-Tuning: A Benchmark](https://mlsanity.com/paper/2402.11592/) (2024)
- [SpikeGPT: Generative Pre-trained Language Model with Spiking Neural Networks](https://mlsanity.com/paper/2302.13939/) (2023)
- [DeepZero: Scaling up Zeroth-Order Optimization for Deep Model Training](https://mlsanity.com/paper/2310.02025/) (2023)
- [A Practical Survey on Faster and Lighter Transformers](https://mlsanity.com/paper/2103.14636/) (2021)
- [Towards Memory- and Time-Efficient Backpropagation for Training Spiking Neural Networks](https://mlsanity.com/paper/2302.14311/) (2023)
- [Efficient On-Chip Learning for Optical Neural Networks Through Power-Aware Sparse Zeroth-Order Optimization](https://mlsanity.com/paper/2012.11148/) (2020)
- [Memory Is All You Need: An Overview of Compute-in-Memory Architectures for Accelerating Large Language Model Inference](https://mlsanity.com/paper/2406.08413/) (2024)
- [ViTA: A Vision Transformer Inference Accelerator for Edge Applications](https://mlsanity.com/paper/2302.09108/) (2023)
- [Spikingformer: A Key Foundation Model for Spiking Neural Networks](https://mlsanity.com/paper/2304.11954/) (2023)
- [Perturbation-efficient Zeroth-order Optimization for Hardware-friendly On-device Training](https://mlsanity.com/paper/2504.20314/) (2025)
- [Memory-Efficient Reversible Spiking Neural Networks](https://mlsanity.com/paper/2312.07922/) (2023)
- [ElasticZO: A Memory-Efficient On-Device Learning with Combined Zeroth- and First-Order Optimization](https://mlsanity.com/paper/2501.04287/) (2025)
- [Xpikeformer: Hybrid Analog-Digital Hardware Acceleration for Spiking Transformers](https://mlsanity.com/paper/2408.08794/) (2024)
- [Stochastic Spiking Attention: Accelerating Attention with Stochastic Computing in Spiking Networks](https://mlsanity.com/paper/2402.09109/) (2024)
- [ASTER: Attention-based Spiking Transformer Engine for Event-driven Reasoning](https://mlsanity.com/paper/2511.06770/) (2025)

## Abstract (arXiv)

Zeroth-order (ZO) optimization estimates gradients using only forward-pass evaluations, making it suitable for fine-tuning non-differentiable, event-driven spiking neural networks (SNNs). However, its deployment on in-memory computing (IMC) accelerators is constrained by the repeated read-modify-write (RMW) operations arising from explicit weight perturbation and the prohibitive hardware footprint of random number generators (RNGs) for statistically independent per-weight perturbations. To address these challenges, we propose an implicit-perturbation ZO (IPZO) architecture in which perturbation sums computed by an event-triggered perturbation generation unit (PGU) are combined with the weighted sums produced by the IMC array, eliminating perturbation-induced RMW operations while preserving weight-stationary execution of IMC. By exploiting spike sparsity, the PGU generates and accumulates perturbation contributions only for spike-activated weight rows, reducing the required row dimension of the RNG array. An address-driven XOR recombination scheme (PGU-XOR) is further introduced to mitigate the spatial correlations caused by direct RNG reuse (PGU-Reuse). The results show that (1) PGU-XOR matches software RNGs in accuracy on Spikingformer/CIFAR-10 (76.41% vs. 76.53%) and perplexity (PPL) on SpikeGPT/WikiText-2 (54.20 vs. 53.23), whereas PGU-Reuse degrades accuracy by 9.56 percentage points and increases PPL by 11.8; (2) implemented in a TSMC 16-nm CMOS technology, PGU-XOR incurs 40.3%-46.0% area and 15.2%-48.9% energy overhead per matrix-vector multiplication relative to PGU-Reuse, yet its faster convergence reduces the total perturbation energy to 0.51x that of PGU-Reuse at iso-accuracy; (3) IPZO reduces the perturbation energy to 0.46x-0.83x that of conventional explicit weight perturbation for a batch size of B=64 and T=4 time steps, with the advantage growing as BT decreases.