Individually Conditional Individual Mutual Information Bound on Generalization Error

被引:8
作者
Zhou, Ruida [1 ]
Tian, Chao [1 ]
Liu, Tie [1 ]
机构
[1] Texas A&M Univ, Dept Elect & Comp Engn, College Stn, TX 77843 USA
基金
美国国家科学基金会;
关键词
Mutual information; Training; Random variables; Heuristic algorithms; Training data; Noise measurement; Upper bound; Information-theoretic bounds; generalization error; stochastic gradient Langevin dynamics;
D O I
10.1109/TIT.2022.3144615
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
We propose an information-theoretic bound on the generalization error based on a combination of the error decomposition technique of Bu et al. and the conditional mutual information (CMI) construction of Steinke and Zakynthinou. In a previous work, Haghifam et al. proposed a different bound combining the two aforementioned techniques, which we refer to as the conditional individual mutual information (CIMI) bound. However, in a simple Gaussian setting, both the CMI and the CIMI bounds are order-wise worse than that by Bu et al. This observation motivated us to propose the bound, which overcomes this issue by reducing the conditioning terms in the conditional mutual information. In the process of establishing this bound, a conditional decoupling lemma is established, which also leads to a meaningful dichotomy and comparison among these information-theoretic bounds. As an application of the proposed bound, we analyze the noisy and iterative stochastic gradient Langevin dynamics and provide an upper bound on its generalization error.
引用
收藏
页码:3304 / 3316
页数:13
相关论文
共 50 条
  • [21] CONDITIONAL DYNAMIC MUTUAL INFORMATION-BASED FEATURE SELECTION
    Liu, Huawen
    Mo, Yuchang
    Zhao, Jianmin
    COMPUTING AND INFORMATICS, 2012, 31 (06) : 1193 - 1216
  • [22] Learning biological network using mutual information and conditional independence
    Dong-Chul Kim
    Xiaoyu Wang
    Chin-Rang Yang
    Jean Gao
    BMC Bioinformatics, 11
  • [23] A Training-Based Mutual Information Lower Bound for Large-Scale Systems
    Gao, Kang
    Meng, Xiangbo
    Laneman, J. Nicholas
    Chisum, Jonathan D.
    Bendlin, Ralf
    Chopra, Aditya
    Hochwald, Bertrand M.
    IEEE TRANSACTIONS ON COMMUNICATIONS, 2022, 70 (08) : 5151 - 5163
  • [24] A Tight Upper Bound on the Mutual Information of Two Boolean Functions
    Pichler, Georg
    Matz, Gerald
    Piantanida, Pablo
    2016 IEEE INFORMATION THEORY WORKSHOP (ITW), 2016,
  • [25] ON THE UTILITY OF CONDITIONAL GENERATION BASED MUTUAL INFORMATION FOR CHARACTERIZING ADVERSARIAL SUBSPACES
    Hsu, Chia-Yi
    Lu, Pei-Hsuan
    Chen, Pin-Yu
    Yu, Chia-Mu
    2018 IEEE GLOBAL CONFERENCE ON SIGNAL AND INFORMATION PROCESSING (GLOBALSIP 2018), 2018, : 1149 - 1153
  • [26] An Improved Feature Selection Algorithm with Conditional Mutual Information for Classification Problems
    Palanichamy, Jaganathan
    Ramasamy, Kuppuchamy
    2013 INTERNATIONAL CONFERENCE ON HUMAN COMPUTER INTERACTIONS (ICHCI), 2013,
  • [27] Conditional and Marginal Mutual Information in Gaussian and Hyperbolic Decay Time Series
    Chavez, Gordon
    JOURNAL OF TIME SERIES ANALYSIS, 2016, 37 (06) : 851 - 861
  • [28] Belavkin-Staszewski Relative Entropy, Conditional Entropy, and Mutual Information
    Zhai, Yuan
    Yang, Bo
    Xi, Zhengjun
    ENTROPY, 2022, 24 (06)
  • [29] How to Gain on Power: Novel Conditional Independence Tests Based on Short Expansion of Conditional Mutual Information
    Kubkowski, Mariusz
    Mielniczuk, Jan
    Teisseyre, Pawel
    JOURNAL OF MACHINE LEARNING RESEARCH, 2021, 22
  • [30] Information-Theoretic Characterizations of Generalization Error for the Gibbs Algorithm
    Aminian, Gholamali
    Bu, Yuheng
    Toni, Laura
    Rodrigues, Miguel R. D.
    Wornell, Gregory W.
    IEEE TRANSACTIONS ON INFORMATION THEORY, 2024, 70 (01) : 632 - 655