.png){kind=icon}
返回
A gradient optimization and manifold preserving based binary neural network for point cloud
.png){kind=icon}
PRE
AI
DOI：10.1016/j.patcog.2023.109445.png)
.png){kind=icon}
原文求助.png){kind=icon}
分享.png){kind=icon}
收藏
摘要
En 中文
With significant progress of deep learning on 3D point cloud, the demand for deployment of point cloud neural network on the edge devices is growing. Binary neural network, a type of quantization compression method, with extreme low bit and fast inference speed, attracts more attention. It is more challenging, but has greater potentiality. Most of the researches on binary networks focus on images rather than point cloud. Considering the particularity of point cloud neural network, this paper presents a novel binarization framework, which includes two main contributions. Firstly, a gradient optimization method is proposed to overcome the shortcomings of Straight Through Estimator (STE) commonly used in the back propagation of binary network training. Secondly, based on the analysis of manifold distortion caused by the binary convolution and pooling operations, we propose an optimized scaling recovery method to restore manifold for the convoluted feature, and also, a pooling correction method to improve the pooled feature's fidelity. Manifold distortion leads to the severe feature homogeneity problem, which brings trouble in generating features with sufficient discrimination for classification and segmentation. The manifold preserving optimizations are designed to introduce minimum extra parameters to balance the accuracy with the computation and storage consumption. Experiments show that the proposed method outperforms state-of-the-art in accuracy with ignored overhead, and also has good scalability.(c) 2023 Elsevier Ltd. All rights reserved.
Keyword:
Binarization
Gradient optimization
Manifold
Point cloud
期刊
IF:
7.6
论文数:
1.3W
被引数:
4.5W
机构
引用论文
M-CBN: Manifold constrained joint image dehazing and super-resolution based on chord boosting network.png)
PATTERN RECOGNITION
PATTERN RECOGNITIONIF7.6
PREAI
分享收藏Spatial information enhancement network for 3D object detection from point cloud基于空间信息增强网络的点云三维目标检测PATTERN RECOGNITION
PATTERN RECOGNITIONIF7.6
PREAI
分享收藏分享收藏Feature selection using bare-bones particle swarm optimization with mutual information基于互信息的裸机粒子群优化算法的特征选择PATTERN RECOGNITION
PATTERN RECOGNITIONIF7.6
PREAI
分享收藏Geometric imbalanced deep learning with feature scaling and boundary sample miningPATTERN RECOGNITION
PATTERN RECOGNITIONIF7.6
PREAI
分享收藏Feature Nonlinear Transformation Non-Negative Matrix Factorization with Kullback-Leibler Divergence具有Kullback-Leibler散度的特征非线性变换非负矩阵分解PATTERN RECOGNITION
PATTERN RECOGNITIONIF7.6
PREAI
分享收藏Environmental Sound Classification on the Edge: A Pipeline for Deep Acoustic Networks on Extremely Resource-Constrained DevicesPATTERN RECOGNITION
OAAI
PATTERN RECOGNITIONIF7.6
OAAI分享收藏From Distortion Manifold to Perceptual Quality: a Data Efficient Blind Image? Quality Assessment ApproachPATTERN RECOGNITIONOAAI
PATTERN RECOGNITIONIF7.6
OAAI分享收藏
查看更多内容