Coincidence detection and integration behavior in spiking neural networks
Bibliographic record
Abstract
Recently, the interest in spiking neural networks (SNNs) remarkably increased, as up to now some key advances of biological neural networks are still out of reach. Thus, the energy efficiency and the ability to dynamically react and adapt to input stimuli as observed in biological neurons is still difficult to achieve. One neuron model commonly used in SNNs is the leaky-integrate-and-fire (LIF) neuron. LIF neurons already show interesting dynamics and can be run in two operation modes: coincidence detectors for low and integrators for high membrane decay times, respectively. However, the emergence of these modes in SNNs and the consequence on network performance and information processing ability is still elusive. In this study, we examine the effect of different decay times in SNNs trained with a surrogate-gradient-based approach. We propose two measures that allow to determine the operation mode of LIF neurons: the number of contributing input spikes and the effective integration interval. We show that coincidence detection is characterized by a low number of input spikes as well as short integration intervals, whereas integration behavior is related to many input spikes over long integration intervals. We find the two measures to linearly correlate via a correlation factor that depends on the decay time. Thus, the correlation factor as function of the decay time shows a powerlaw behavior, which could be an intrinsic property of LIF networks. We argue that our work could be a starting point to further explore the operation modes in SNNs to boost efficiency and biological plausibility. Supplementary Information: The online version of this article (10.1007/s11571-023-10038-0) contains supplementary material, which is available to authorized users.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.010 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".