Efficient 3D Recognition with Event-driven Spike Sparse Convolution

X Qiu, M Yao, J Zhang, Y Chou, N Qiao, S Zhou… - arxiv preprint arxiv …, 2024 - arxiv.org
Spiking Neural Networks (SNNs) provide an energy-efficient way to extract 3D spatio-
temporal features. Point clouds are sparse 3D spatial data, which suggests that SNNs …

Quantized Spike-driven Transformer

X Qiu, J Zhang, W Wei, H Cao, J Guo, RJ Zhu… - arxiv preprint arxiv …, 2025 - arxiv.org
Spiking neural networks are emerging as a promising energy-efficient alternative to
traditional artificial neural networks due to their spike-driven paradigm. However, recent …

Universal Image Restoration Pre-training via Degradation Classification

JK Hu, L **, Z Yao, Y Lu - arxiv preprint arxiv:2501.15510, 2025 - arxiv.org
This paper proposes the Degradation Classification Pre-Training (DCPT), which enables
models to learn how to classify the degradation type of input images for universal image …

Beyond Timesteps: A Novel Activation-wise Membrane Potential Propagation Mechanism for Spiking Neural Networks in 3D cloud

J Song, B Zheng, X Yang, D Wang - arxiv preprint arxiv:2502.12791, 2025 - arxiv.org
Due to the similar characteristics between event-based visual data and point clouds, recent
studies have emerged that treat event data as event clouds to learn based on point cloud …