architecture · first seen 7 Oct 2026
depthwise separable convolution
Also: depth-wise separable convolutions
Convolution factorized into per-channel spatial filtering followed by pointwise mixing, reducing parameters and compute.
Mentions per week
PapersCodeDiscussion
Earliest sources
- Efficient 3D Gaussian Head Avatars for Edge DevicesPapers · 7 Oct 2026 · Umar Farooq, Jean-Yves Guillemaut, Adrian Hilton, Marco Volino
- Sea-Net: A Small and Inherently Interpretable Neural Network for Time Series ClassificationPapers · 13 Oct 2026 · Ubaid Ur Rehman, Tanmoy Mondal
- ConvLUT: Rethinking Convolutional Architectures for Ultra-Low-Latency LUT-Based InferencePapers · 8 Nov 2026 · Marta Andronic, Olivia Weng, Danial Zuberi, Ryan Kastner, George A. Constantinid…
- A Compact xDAWN-Guided Spatio-Temporal Network with Filter-Bank Convolution for P300 DetectionPapers · 12 Nov 2026 · Mohammad Mahdi Habibi Bina, Mohammad Reza Daliri
Latest sources
- A Compact xDAWN-Guided Spatio-Temporal Network with Filter-Bank Convolution for P300 DetectionPapers · 12 Nov 2026 · Mohammad Mahdi Habibi Bina, Mohammad Reza Daliri
- ConvLUT: Rethinking Convolutional Architectures for Ultra-Low-Latency LUT-Based InferencePapers · 8 Nov 2026 · Marta Andronic, Olivia Weng, Danial Zuberi, Ryan Kastner, George A. Constantinid…
- Sea-Net: A Small and Inherently Interpretable Neural Network for Time Series ClassificationPapers · 13 Oct 2026 · Ubaid Ur Rehman, Tanmoy Mondal
- Efficient 3D Gaussian Head Avatars for Edge DevicesPapers · 7 Oct 2026 · Umar Farooq, Jean-Yves Guillemaut, Adrian Hilton, Marco Volino