Jul 20, 2020 · In this paper, we present a sparsity-inducing regularization term based on the ratio l1/l2 pseudo-norm defined on the filter coefficients.

Dec 9, 2020 · We therefore propose a new strategy, based on l1/l2-norm, to obtain a subset of kernels with all weights equal to zero (such as that the ...

This paper presents a sparsity-inducing regularization term based on the ratio l1/l2 pseudo-norm defined on the filter coefficients, which significantly ...

Jan 11, 2021 · Ba- seline is the simple LeNet Caffe model. l1 and l2 are the best results found by using the l1-norm and l2-norm regularization on the kernels.

Sep 6, 2024 · By defining this pseudo-norm appropriately for the different filter kernels, and removing irrelevant filters, the number of kernels in each ...

Learning Sparse Convolutional Neural Network via Quantization ... Learning Sparse Filters in Deep Convolutional Neural Networks with a l1/l2 Pseudo-Norm.

This paper proposes fast general normalized convolutional sparse filtering (FGNC-SF) via the L 1 -L 2 mixed norm for intelligent fault diagnosis.

Nov 24, 2019 · In this paper, we present a sparsity-inducing regularization term based on the ratio l1/l2 pseudo-norm defined on the filter coefficients. By ...

Learning Sparse Filters in Deep Convolutional Neural Networks with a l1/l2 Pseudo-Norm. July 2020. Anthony Berthelier · Yongzhe ...

Thus, we also propose an efficient sparse matrix multiplication algorithm. Based on the fact that the sparse convolutional kernels are fixed after training, we ...