Curvature-based Comparison of Two Neural Networks

被引:0
|
作者
Yu, Tao [1 ]
Long, Huan [2 ]
Hopcroft, John E. [3 ]
机构
[1] Shanghai Jiao Tong Univ, Dept Math, Shanghai, Peoples R China
[2] Shanghai Jiao Tong Univ, Dept Comp Sci & Engn, Shanghai, Peoples R China
[3] Cornell Uniers, Dept Comp Sci, New York, NY USA
来源
2018 24TH INTERNATIONAL CONFERENCE ON PATTERN RECOGNITION (ICPR) | 2018年
关键词
D O I
暂无
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
In this paper we show the similarities and differences of two deep neural networks by comparing the manifolds composed of activation vectors in each fully connected layer of them. The main contribution of this paper includes 1) a new data generating algorithm which is crucial for determining the dimension of manifolds; 2) a systematic strategy to compare manifolds. Especially, we take Riemann curvature and sectional curvature as part of criterion, which can reflect the intrinsic geometric properties of manifolds. Some interesting results and phenomenon are given, which help in specifying the similarities and differences between the features extracted by two networks and demystifying the intrinsic mechanism of deep neural networks.
引用
收藏
页码:441 / 447
页数:7
相关论文
共 50 条
  • [1] Curvature-Based Pooling Within Graph Neural Networks
    Sanders, Cedric
    Roth, Andreas
    Liebig, Thomas
    MACHINE LEARNING AND PRINCIPLES AND PRACTICE OF KNOWLEDGE DISCOVERY IN DATABASES, ECML PKDD 2023, PT III, 2025, 2135 : 471 - 485
  • [2] CurvAGN: Curvature-based Adaptive Graph Neural Networks for Predicting Protein-Ligand Binding Affinity
    Wu, Jianqiu
    Chen, Hongyang
    Cheng, Minhao
    Xiong, Haoyi
    BMC BIOINFORMATICS, 2023, 24 (01)
  • [3] CurvAGN: Curvature-based Adaptive Graph Neural Networks for Predicting Protein-Ligand Binding Affinity
    Jianqiu Wu
    Hongyang Chen
    Minhao Cheng
    Haoyi Xiong
    BMC Bioinformatics, 24
  • [4] Curvature-based stroke rendering
    Saito, Suguru
    Kani, Akane
    Chang, Youngha
    Nakajima, Masayuki
    VISUAL COMPUTER, 2008, 24 (01): : 1 - 11
  • [5] Curvature-based surface regeneration
    Robinson, Sebastian T.
    Mullineux, Glen
    MATHEMATICS OF SURFACES XII, PROCEEDINGS, 2007, 4647 : 378 - +
  • [6] Curvature-based stroke rendering
    Suguru Saito
    Akane Kani
    Youngha Chang
    Masayuki Nakajima
    The Visual Computer, 2008, 24 : 1 - 11
  • [7] Two new preconditioners for mean curvature-based image deblurring problem
    Ahmad, Shahbaz
    Al-Mahdi, Adel M.
    Ahmed, Rashad
    AIMS MATHEMATICS, 2021, 6 (12): : 13824 - 13844
  • [8] Two-Dimensional Curvature-Based Analysis of Intrinsic Mode Functions
    Diop, El Hadji S.
    Alexandre, Radjesvarane
    Boudra, Abdel-O.
    IEEE SIGNAL PROCESSING LETTERS, 2018, 25 (01) : 20 - 24
  • [9] Curvature-based fuzzy surface classification
    Ramalingam, Soodamani
    Liu, Zhi-Qiang
    Iourinski, Dmitri
    IEEE TRANSACTIONS ON FUZZY SYSTEMS, 2006, 14 (04) : 573 - 589
  • [10] A CURVATURE-BASED APPROACH TO TERRAIN RECOGNITION
    GOLDGOF, DB
    HUANG, TS
    HUA, L
    IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 1989, 11 (11) : 1213 - 1217