Algorithms Noise Robustness in SNNs Spiking neural networks tolerate synaptic noise far better than ANNs thanks to temporal spike integration acting as a natural noise filter. Noise Exploitation for ANN-to-SNN Conversion Injecting controllable noise during ANN quantization training to model SNN spike dynamics, enabling seamless end-to-end ANN-to-SNN conversion. Low-Latency Spiking Neural Networks QFFS quantization framework and dynamic confidence decoding for ultra-low-latency SNN inference on ImageNet. Devices & Hardware Skyrmionic Synapses Nanoscale multilayer skyrmion synapses operating at room temperature for deep spiking neural networks. Analog In-Memory Computing NPU Heterogeneous embedded NPU combining PCM-based analog in-memory computing tiles with digital accelerators for efficient neural network inference. AHWA-LoRA for Analog In-Memory Computing Hardware-aware low-rank adaptation training that keeps pre-trained transformer weights fixed on analog AIMC tiles while training lightweight LoRA adapters.
