Journal of Computer Applications ›› 2023, Vol. 43 ›› Issue (5): 1401-1408.DOI: 10.11772/j.issn.1001-9081.2022040581

• Artificial intelligence • Previous Articles     Next Articles

Adaptive multi-scale feature channel grouping optimization algorithm based on NSGA‑Ⅱ

Bin WANG1,2(), Tian XIANG1, Yidong LYU1, Xiaofan WANG1   

  1. 1.School of Computer Science and Engineering,Xi'an University of Technology,Xi'an Shaanxi 710048,China
    2.Shaanxi Key Laboratory for Network Computing and Security Technology (Xi'an University of Technology),Xi'an Shaanxi 710048,China
  • Received:2022-04-24 Revised:2022-06-19 Accepted:2022-07-01 Online:2022-07-26 Published:2023-05-10
  • Contact: Bin WANG
  • About author:WANG Bin, born in 1971, M. S., associate professor. His research interests include evolutionary computing, neural network.
    XIANG Tian, born in 1997, M. S. candidate. Her research interests include evolutionary computing.
    LYU Yidong, born in 1995, M. S., engineer. His research interests include neural network.
    WANG Xiaofan, born in 1976, Ph. D., professor. His research interests include intelligent information processing.
  • Supported by:
    National Natural Science Foundation of China(61976177)

基于NSGA‑Ⅱ的自适应多尺度特征通道分组优化算法

王彬1,2(), 向甜1, 吕艺东1, 王晓帆1   

  1. 1.西安理工大学 计算机科学与工程学院,西安 710048
    2.陕西省网络计算与安全技术重点实验室(西安理工大学),西安 710048
  • 通讯作者: 王彬
  • 作者简介:王彬(1971—),男,陕西西安人,副教授,硕士,CCF会员,主要研究方向:进化计算、神经网络 wb@xaut.edu.cn
    向甜(1997—),女,陕西西安人,硕士研究生,主要研究方向:进化计算
    吕艺东(1995—),男,陕西西安人,工程师,硕士,主要研究方向:神经网络
    王晓帆(1976—),男,河北石家庄人,教授,博士,CCF会员,主要研究方向:智能信息处理。
  • 基金资助:
    国家自然科学基金资助项目(61976177)

Abstract:

Aiming at the balance optimization problem of Lightweight Convolutional Neural Network (LCNN) in accuracy and complexity, an adaptive multi-scale feature channel grouping optimization algorithm based on fast Non-dominated Sorting Genetic Algorithm (NSGA-Ⅱ) was proposed to optimize the feature channel grouping structure of LCNN. Firstly, the complexity minimization and accuracy maximization of the feature fusion layer structure in LCNN were regarded as two optimization objectives, and the dual-objective function modeling and theoretical analysis were carried out. Then, a LCNN structure optimization framework based on NSGA-Ⅱ was designed, and an adaptive grouping layer based on NSGA-Ⅱ was added to deep convolution layer in original LCNN structure, thus constructing an Adaptive Multi-scale Feature Fusion Network based on NSGA2 (NSGA2-AMFFNetwork). Experimental results on image classification datasets show that compared with the manually designed network structure M_blockNet_v1, NSGA2-AMFFNetwork has the average accuracy improved by 1.220 2 percentage points, and the running time decreased by 41.07%. This above indicates that the proposed optimization algorithm can balance the complexity and accuracy of LCNN, and also provide more options for network structure with balanced performance for ordinary users who lack domain knowledge.

Key words: Lightweight Convolutional Neural Network (LCNN), feature extraction channel grouping optimization, dual-objective function modeling, fast Non-dominated Sorting Genetic Algorithm (NSGA?Ⅱ), image classification, evolutionary algorithm

摘要:

针对轻量型卷积神经网络(LCNN)的精确度和复杂度均衡优化问题,提出基于快速非支配排序遗传算法(NSGA-Ⅱ)的自适应多尺度特征通道分组优化算法对LCNN特征通道分组结构进行优化。首先,将LCNN中的特征融合层结构的复杂度最小化和精确度最大化作为两个优化目标,进行双目标函数建模及理论分析;然后,设计基于NSGA-Ⅱ的LCNN结构优化框架,并在原始LCNN结构的深度卷积层之上增加基于NSGA-Ⅱ的自适应分组层,构建基于NSGA-Ⅱ的自适应多尺度的特征融合网络NSGA2-AMFFNetwork。在图像分类数据集上的实验结果显示,与手工设计的网络结构M_blockNet_v1相比,NSGA2-AMFFNetwork的平均精确度提升了1.220 2个百分点,运行时间降低了41.07%。这表明所提优化算法能较好平衡LCNN的复杂度和精确度,同时还可为领域知识不足的普通用户提供更多性能表现均衡的网络结构选择方案。

关键词: 轻量型卷积神经网络, 特征提取通道分组优化, 双目标函数建模, 快速非支配排序遗传算法, 图像分类, 进化算法

CLC Number: