![[2411.04760] Zero-Shot Temporal Resolution Domain Adaptation for Spiking Neural Networks](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2411.04760] Zero-Shot Temporal Resolution Domain Adaptation for Spiking Neural Networks
Summary
This paper presents novel domain adaptation methods for Spiking Neural Networks (SNNs) to address performance drops due to mismatched temporal resolutions between training and deployment data.
Why It Matters
As SNNs gain traction for their energy efficiency in neuromorphic computing, adapting them to varying temporal resolutions is crucial for their practical deployment. This research offers innovative solutions that enhance SNN performance without the need for retraining, making them more versatile in real-world applications.
Key Takeaways
- Proposes three novel methods for zero-shot domain adaptation in SNNs.
- Demonstrates significant performance improvements over existing methods.
- Highlights the ability to achieve high accuracy on high temporal resolution data using lower resolution training.
- Evaluates methods on audio keyword spotting and neuromorphic image datasets.
- Addresses a critical challenge in deploying SNNs in diverse environments.
Computer Science > Machine Learning arXiv:2411.04760 (cs) [Submitted on 7 Nov 2024 (v1), last revised 18 Feb 2026 (this version, v3)] Title:Zero-Shot Temporal Resolution Domain Adaptation for Spiking Neural Networks Authors:Sanja Karilanova, Maxime Fabre, Emre Neftci, Ayça Özçelikkale View a PDF of the paper titled Zero-Shot Temporal Resolution Domain Adaptation for Spiking Neural Networks, by Sanja Karilanova and 3 other authors View PDF HTML (experimental) Abstract:Spiking Neural Networks (SNNs) are biologically-inspired deep neural networks that efficiently extract temporal information while offering promising gains in terms of energy efficiency and latency when deployed on neuromorphic devices. SNN parameters are sensitive to temporal resolution, leading to significant performance drops when the temporal resolution of target data during deployment is not the same as that of the source data used for training, especially when fine-tuning with the target data is not possible during deployment. To address this challenge, we propose three novel domain adaptation methods for adapting neuron parameters to account for the change in time resolution without re-training on target time resolution. The proposed methods are based on a mapping between neuron dynamics in SNNs and State Space Models (SSMs) and are applicable to general neuron models. We evaluate the proposed methods under spatio-temporal data tasks, namely the audio keyword spotting datasets SHD and MSWC, and the neuro...