Sparse Coding and Temporal Pattern Learning Co-Mediated by Dual Spike-Timing-Dependent Plasticity in a Multilayer Excitatory–Inhibitory Spiking Network
Biomimetics · 02.07.2026 · 11 (7) · 462
DOI: 10.3390/biomimetics11070462 ↗Kurz gefasst
Fachartikel in der Zeitschrift Biomimetics, online erschienen am 02.07.2026. Verfasst von Chunhua Yuan und 3 weiteren Autor:innen. Beteiligte Einrichtung: School of Automation and Electrical Engineering, Shenyang Ligong University, Shenyang 110159, China. Die Arbeit nennt 32 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: National Natural Science Foundation of China, Scientific Research Fund of Liaoning Provincial Education Department.
Abstract (Originalsprache)
Excitatory–inhibitory (E-I) local circuits play a central role in synaptic plasticity and neural coding, yet their multilayer learning dynamics remain poorly understood. We constructed a multilayer feedforward spiking neural network with intra-layer E-I connectivity, using Izhikevich neurons to model regular spiking (RS) and fast spiking (FS) cells, and examined cooperative learning under excitatory and inhibitory spike-timing-dependent plasticity (eSTDP and iSTDP). FS-mediated lateral inhibition alleviates the long-term depression bias arising from RS firing rate adaptation via winner-take-all competition, promoting heterogeneous E→E weight differentiation while preserving mean synaptic strength. A 12×12 parameter grid scan shows that iSTDP expands the stable learning region in the E-I parameter space and reveals a sustained cooperative co-evolution of eSTDP and iSTDP during training. For sparse coding, RS adaptation is the primary driver of Lifetime Sparseness, with FS inhibition acting as a cooperative enhancer; the network exhibits low sparseness at the input layer, a rapid increase at the second layer, and a stable plateau in deeper layers. For temporal pattern learning, the selectivity index d′ improved substantially after training, reaching approximately 1.90 times that of the FS-absent condition; both interval sensitivity and pattern generalization tests confirmed that this advantage is robust across biologically plausible inter-group delays and preserved under small temporal jitter. Mutual information analysis reveals a consistent tendency for intra-layer FS circuits to maintain higher stimulus-related information across deep layers, consistent with FS-mediated suppression of non-specific responses. These findings provide computational evidence, within the scope of the present model, for understanding cortical E-I cooperative plasticity and inform design principles for neuromorphic systems with adaptive inhibitory regulation.
Schlagwörter
Stichwörter der Autor:innen
Autor:innen und Institutionen
- School of Automation and Electrical Engineering, Shenyang Ligong University, Shenyang 110159, China.
- College of Information Science and Engineering, Northeastern University, Shenyang 110819, China.
ORCID-Profile: Chunhua Yuan
Förderung
- National Natural Science Foundation of China
- Scientific Research Fund of Liaoning Provincial Education Department
Zitieren
APA
Yuan, C., Wang, D., Li, X., & Gao, X. (2026). Sparse Coding and Temporal Pattern Learning Co-Mediated by Dual Spike-Timing-Dependent Plasticity in a Multilayer Excitatory–Inhibitory Spiking Network. Biomimetics, 11(7), 462. https://doi.org/10.3390/biomimetics11070462
Vancouver
Yuan C, Wang D, Li X, Gao X. Sparse Coding and Temporal Pattern Learning Co-Mediated by Dual Spike-Timing-Dependent Plasticity in a Multilayer Excitatory–Inhibitory Spiking Network. Biomimetics. 2026;11(7):462. doi:10.3390/biomimetics11070462
BibTeX
@article{Yuan2026,
author = {Yuan, Chunhua and Wang, Deyang and Li, Xiangyu and Gao, Xianwen},
title = {Sparse Coding and Temporal Pattern Learning Co-Mediated by Dual Spike-Timing-Dependent Plasticity in a Multilayer Excitatory–Inhibitory Spiking Network},
journal = {Biomimetics},
year = {2026},
volume = {11},
number = {7},
pages = {462},
doi = {10.3390/biomimetics11070462}
}
EUVIMED