Research map / Architectures and efficiency / Spiking networks
Beyond Low-Pass Dynamics: Frequency-Selective Spiking Reservoirs with Resonant Neurons
Aurora Micheli, Claudio Gallicchio, Nergis Tomen
Read the paperCheck your idea against 137,217 papers
AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Proposes frequency-selective spiking reservoirs built from resonant spiking neurons instead of standard LIF neurons, to overcome the low-pass limitation of existing spiking reservoir computing.
Problem
Existing spiking reservoir computing almost exclusively uses LIF neurons, whose low-pass dynamics limit their ability to distinguish information in different frequency bands.
Approach
Replace LIF neurons in spiking reservoirs with resonant neurons so the reservoir becomes frequency-selective, retaining event-driven sparse spiking while gaining band-selective dynamics. Full method details are unavailable beyond the abstract.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| replace heterogeneous LIF neurons with heterogeneous HRF neurons (ablation) | spiking reservoir (LIF-Res) · sMNIST (N=800) | accuracy (%): 67.00 -> 93.64 | Improved |
| replace LIF neurons with HRF neurons (ablation) | spiking reservoir (LIF-Res) · SHD (N=3,000) | accuracy (%): 72.95 -> 83.48 | Improved |
| replace LIF neurons with HRF neurons (ablation) | spiking reservoir (LIF-Res) · FordA (N=800) | accuracy (%): 67.15 -> 69.72 | Improved |
| replace LIF neurons with HRF neurons (ablation) | spiking reservoir (LIF-Res) · DVS Gesture (N=3,000) | accuracy (%): 73.48 -> 71.81 | No clear effect |
| replace LIF neurons with HRF neurons (ablation) | spiking reservoir (LIF-Res) · FordA (N=800) | CKA class discriminability: +322% | Improved |
| replace LIF neurons with HRF neurons (ablation) | spiking reservoir (LIF-Res) · sMNIST (N=800) | CKA class discriminability: +9% | No clear effect |
| replace LIF neurons with HRF neurons (ablation) | spiking reservoir (LIF-Res) · sMNIST, FordA, SHD, DVS Gesture | fraction of resonant neurons (Q > 0): 98%, 100%, 62%, 83% vs 23%, 9%, 15%, 4% | Improved |
| replace LIF neurons with HRF neurons (ablation) | spiking reservoir (LIF-Res) · sMNIST (N=800) | energy (x10^-4 J): 2.352 -> 1.676 | Improved |
| replace LIF neurons with HRF neurons (ablation) | spiking reservoir (LIF-Res) · DVS Gesture (N=3,000) | energy (x10^-3 J): 0.953 -> 1.265 | Hurt |
| use spike-based mean firing-rate readout instead of best membrane-based readout (ablation) | HRF-Res · sMNIST, FordA, SHD, DVS Gesture | accuracy (%) | Hurt |
| use temporal mean readout instead of final state (ablation) | HRF-Res · FordA | accuracy (%): ~4% | Improved |
| use combined 3N readout (RMS + Std + final state) (ablation) | HRF-Res · sMNIST, FordA, SHD, DVS Gesture | accuracy (%) | Improved |
Experimental setup
Baselines: LIF-based spiking reservoirs
Builds on: leaky integrate-and-fire neuron spiking reservoir computing
Closest papers by meaning
- Balanced Resonate-and-Fire Neurons (ICML 2024 · similarity 0.88)
- FiTS: Interpretable Spiking Neurons via Frequency Selectivity and Temporal Shaping (NeurIPS 2026 · similarity 0.87)
- Spiking Neural Networks Need High-Frequency Information (NeurIPS 2025 · similarity 0.87)
- SpikingSSMs: Learning Long Sequences with Sparse and Parallel Spiking State Space Models (AAAI 2025 · similarity 0.86)
- Frequency Domain Reservoir Computing (NeurIPS 2026 · similarity 0.86)
- The Dormant Spiking Neuron: A State-Driven Mechanism for Efficient Spiking Neural Networks (NeurIPS 2026 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
