Towards High-performance Spiking Transformers from ANN to SNN Conversion
Spiking neural networks (SNNs) show great potential due to their energy efficiency, fast
processing capabilities, and robustness. There are two main approaches to constructing …
processing capabilities, and robustness. There are two main approaches to constructing …
Hybrid neural networks for continual learning inspired by corticohippocampal circuits
Q Shi, F Liu, H Li, G Li, L Shi, R Zhao - Nature Communications, 2025 - nature.com
Current artificial systems suffer from catastrophic forgetting during continual learning, a
limitation absent in biological systems. Biological mechanisms leverage the dual …
limitation absent in biological systems. Biological mechanisms leverage the dual …
Brain-inspired continual pre-trained learner via silent synaptic consolidation
Pre-trained models have demonstrated impressive generalization capabilities, yet they
remain vulnerable to catastrophic forgetting when incrementally trained on new tasks …
remain vulnerable to catastrophic forgetting when incrementally trained on new tasks …
Enhancing Generalization and Convergence in Neural Networks through a Dual-Phase Regularization Approach with Excitatory-Inhibitory Transition
M Xu, H Yin, S Zhong - 2024 International Conference on …, 2024 - ieeexplore.ieee.org
Overfitting and slow convergence represent common hurdles encountered during the
training of deep neural networks (DNNs). While conventional regularization methods like …
training of deep neural networks (DNNs). While conventional regularization methods like …