{"paper":{"title":"NeuroTrain: Surveying Local Learning Rules for Spiking Neural Networks with an Open Benchmarking Framework","license":"http://creativecommons.org/licenses/by/4.0/","headline":"A single taxonomy sorts spiking neural network training methods by their signals and locality while a shared code base lets researchers test them together.","cross_cats":["cs.AI"],"primary_cat":"cs.NE","authors_text":"Alessandro Savino, Alessio Caviglia, Filippo Marostica, Roberta Bardini, Stefano Di Carlo","submitted_at":"2026-05-14T16:50:15Z","abstract_excerpt":"The rapid expansion of spiking neural networks (SNNs) has led to a proliferation of training algorithms that differ widely in biological inspiration, computational structure, and hardware suitability. Despite this progress, the field lacks a unified, fine-grained taxonomy that systematically organizes these approaches and clarifies their conceptual relationships. This survey provides a comprehensive taxonomy of SNN training algorithms, spanning surrogate-gradient backpropagation, local and three-factor learning rules, biologically inspired plasticity mechanisms, ANN-to-SNN conversion pipelines"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"The survey provides a comprehensive taxonomy of SNN training algorithms spanning surrogate-gradient backpropagation, local and three-factor learning rules, biologically inspired plasticity, ANN-to-SNN conversion, and non-standard optimization, supported by the release of NeuroTrain for consistent benchmarking.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the representative algorithms implemented in NeuroTrain sufficiently capture the diversity and key properties of the broader literature without significant omissions or implementation biases.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"A taxonomy of SNN training algorithms is presented with the release of NeuroTrain, an open benchmarking framework for reproducible comparisons across datasets and architectures.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"A single taxonomy sorts spiking neural network training methods by their signals and locality while a shared code base lets researchers test them together.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"d25edcc6fccd85dbcc11ba11551f6b9b58cb4b53031a51c144b848849b88d2fa"},"source":{"id":"2605.15058","kind":"arxiv","version":1},"verdict":{"id":"224c3af4-49c6-4479-9886-ef8a778c1e4a","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-15T03:05:56.906634Z","strongest_claim":"The survey provides a comprehensive taxonomy of SNN training algorithms spanning surrogate-gradient backpropagation, local and three-factor learning rules, biologically inspired plasticity, ANN-to-SNN conversion, and non-standard optimization, supported by the release of NeuroTrain for consistent benchmarking.","one_line_summary":"A taxonomy of SNN training algorithms is presented with the release of NeuroTrain, an open benchmarking framework for reproducible comparisons across datasets and architectures.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the representative algorithms implemented in NeuroTrain sufficiently capture the diversity and key properties of the broader literature without significant omissions or implementation biases.","pith_extraction_headline":"A single taxonomy sorts spiking neural network training methods by their signals and locality while a shared code base lets researchers test them together."},"references":{"count":191,"sample":[{"doi":"","year":1997,"title":"Maass, Networks of spiking neurons: The third generation of neural network models, Neural Net- works 10 (9) (1997) 1659–1671.doi:10.1016/ S0893-6080(97)00011-7","work_id":"f9c322e9-cc35-4211-b971-7dd7265b6cb5","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"10.1109/mm.2018","year":2018,"title":"M. Davies, N. Srinivasa, T.-H. Lin, G. Chinya, Y . Cao, S. H. Choday, G. Dimou, P. Joshi, N. Imam, S. Jain, Y . Liao, C.-K. Lin, A. Lines, R. Liu, D. Mathaikutty, S. McCoy, A. Paul, J. Tse, G. Venkata","work_id":"9b6fde68-321e-4f9b-8f3a-6c06f85e761b","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"10.1109/access.2025.3566717","year":2025,"title":"B. Huo, F. Li, S. Peng, H. Chen, S. Xin, H. Wang, Research on SNN Learning Algorithms and Networks Based on Biological Plausibility, IEEE Access 13 (2025) 95243–95256.doi:10.1109/ACCESS.2025.3566717","work_id":"073c7bbe-83c4-4882-be82-1e6bba14e580","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"10.1007/s00422-008-0237-x","year":2008,"title":"P. Lansky, S. Ditlevsen, A review of the methods for signal estimation in stochastic diffusion leaky integrate- and-fire neuronal models, Biological cybernetics 99 (4) (2008) 253–262.doi:10.1007/s0042","work_id":"39e53955-83fe-4d31-9b11-d5a8507a6954","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2025,"title":"B. Mészáros, J. C. Knight, T. Nowotny, Efficient event- based delay learning in spiking neural networks, Nature Communications 16 (1) (2025) 10422.doi:10.1038/ s41467-025-65394-8","work_id":"3121ec1a-8024-4525-b10b-a7a9c925f07e","ref_index":6,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":191,"snapshot_sha256":"2db4a6c86d8313cb7a76ff232913296877c124acd05938c1e6e5aeb1d2536cb7","internal_anchors":1},"formal_canon":{"evidence_count":2,"snapshot_sha256":"8d7cbe5f335559ad18688527134f727d70ec23dc538a3823d18e255808eddf9b"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}