Trapezoidal Step Scheduler for Model-Agnostic Meta-Learning in Medical Imaging

被引:0
|
作者
Voon, Wingates [1 ]
Hum, Yan Chai [1 ]
Tee, Yee Kai [1 ]
Yap, Wun-She [2 ]
Lai, Khin Wee [3 ]
Nisar, Humaira [4 ]
Mokayed, Hamam [5 ]
机构
[1] Univ Tunku Abdul Rahman, Lee Kong Chian Fac Engn & Sci, Dept Mechatron & Biomed Engn, Sungai Long, Malaysia
[2] Univ Tunku Abdul Rahman, Lee Kong Chian Fac Engn & Sci, Dept Elect & Elect Engn, Sungai Long, Malaysia
[3] Univ Malaya, Dept Biomed Engn, Kuala Lumpur, Malaysia
[4] Univ Tunku Abdul Rahman, Fac Engn & Green Technol, Dept Elect Engn, Kampar, Malaysia
[5] Lulea Univ Technol, Dept Comp Sci Elect & Space Engn, Lulea, Sweden
关键词
Few-shot learning; Medical image classification; Trapezoidal step scheduler; Model-agnostic meta-learning;
D O I
10.1016/j.patcog.2024.111316
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Model-Agnostic Meta-learning (MAML) is a widely adopted few-shot learning (FSL) method designed to mitigate the dependency on large, labeled datasets of deep learning-based methods in medical imaging analysis. However, MAML's reliance on a fixed number of gradient descent (GD) steps for task adaptation results in computational inefficiency and task-level overfitting. To address this issue, we introduce Tra-MAML, which optimizes the balance between model adaptation capacity and computational efficiency through a trapezoidal step scheduler (TRA). The TRA scheduler dynamically adjusts the number of GD steps in the inner optimization loop: initially increasing the steps uniformly to reduce variance, maintaining the maximum number of steps to enhance adaptation capacity, and finally decreasing the steps uniformly to mitigate overfitting. Our evaluation of TraMAML against selected FSL methods across four medical imaging datasets demonstrates its superior performance. Notably, Tra-MAML outperforms MAML by 13.36% on the BreaKHis40X dataset in the 3-way 10-shot scenario.
引用
收藏
页数:16
相关论文
共 50 条
  • [41] Kronecker-factored Approximate Curvature with adaptive learning rate for optimizing model-agnostic meta-learning
    Zhang, Ce
    Yao, Xiao
    Shi, Changfeng
    Gu, Min
    MULTIMEDIA SYSTEMS, 2023,
  • [42] Convolutional Shrinkage Neural Networks Based Model-Agnostic Meta-Learning for Few-Shot Learning
    Yunpeng He
    Chuanzhi Zang
    Peng Zeng
    Qingwei Dong
    Ding Liu
    Yuqi Liu
    Neural Processing Letters, 2023, 55 : 505 - 518
  • [43] Fast Adaptive Active Noise Control Based on Modified Model-Agnostic Meta-Learning Algorithm
    Shi, Dongyuan
    Gan, Woon-Seng
    Lam, Bhan
    Ooi, Kenneth
    IEEE SIGNAL PROCESSING LETTERS, 2021, 28 : 593 - 597
  • [44] Model-Agnostic Meta-Learning for EEG Motor Imagery Decoding in Brain-Computer-Interfacing
    Li, Denghao
    Ortega, Pablo
    Wei, Xiaoxi
    Faisal, Aldo
    2021 10TH INTERNATIONAL IEEE/EMBS CONFERENCE ON NEURAL ENGINEERING (NER), 2021, : 527 - 530
  • [45] Kronecker-factored Approximate Curvature with adaptive learning rate for optimizing model-agnostic meta-learning
    Zhang, Ce
    Yao, Xiao
    Shi, Changfeng
    Gu, Min
    MULTIMEDIA SYSTEMS, 2023, 29 (06) : 3169 - 3177
  • [46] Convolutional Shrinkage Neural Networks Based Model-Agnostic Meta-Learning for Few-Shot Learning
    He, Yunpeng
    Zang, Chuanzhi
    Zeng, Peng
    Dong, Qingwei
    Liu, Ding
    Liu, Yuqi
    NEURAL PROCESSING LETTERS, 2023, 55 (01) : 505 - 518
  • [47] Learning From Visual Demonstrations via Replayed Task-Contrastive Model-Agnostic Meta-Learning
    Hu, Ziye
    Li, Wei
    Gan, Zhongxue
    Guo, Weikun
    Zhu, Jiwei
    Wen, James Zhiqing
    Zhou, Decheng
    IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY, 2022, 32 (12) : 8756 - 8767
  • [48] Kronecker-factored Approximate Curvature with adaptive learning rate for optimizing model-agnostic meta-learning
    Ce Zhang
    Xiao Yao
    Changfeng Shi
    Min Gu
    Multimedia Systems, 2023, 29 (6) : 3169 - 3177
  • [49] Few-Shot Class-Adaptive Anomaly Detection with Model-Agnostic Meta-Learning
    Feng, Tongtong
    Qi, Qi
    Wang, Jingyu
    Liao, Jianxin
    2021 IFIP NETWORKING CONFERENCE AND WORKSHOPS (IFIP NETWORKING), 2021,
  • [50] Combating Out-of-distribution Errors using Model-Agnostic Meta-Learning for Digital Pathology
    Fagerblom, Freja
    Stacke, Karin
    Molin, Jesper
    MEDICAL IMAGING 2021 - DIGITAL PATHOLOGY, 2021, 11603