Information theoretic limits of learning a sparse rule

被引:0
|
作者
Luneau, Clement [1 ]
Macris, Nicolas [1 ]
Barbier, Jean [2 ]
机构
[1] Ecole Polytech Fed Lausanne, Lausanne, Switzerland
[2] Abdus Salaam Int Ctr Theoret Phys, Trieste, Italy
关键词
MUTUAL INFORMATION; TIGHT BOUNDS; SHARP BOUNDS; CAPACITY; ERROR;
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
We consider generalized linear models in regimes where the number of nonzero components of the signal and accessible data points are sublinear with respect to the size of the signal. We prove a variational formula for the asymptotic mutual information per sample when the system size grows to infinity. This result allows us to derive an expression for the minimum mean-square error (MMSE) of the Bayesian estimator when the signal entries have a discrete distribution with finite support. We find that, for such signals and suitable vanishing scalings of the sparsity and sampling rate, the MMSE is nonincreasing piecewise constant. In specific instances the MMSE even displays an all-or-nothing phase transition, that is, the MMSE sharply jumps from its maximum value to zero at a critical sampling rate. The all-or-nothing phenomenon has previously been shown to occur in high-dimensional linear regression. Our analysis goes beyond the linear case and applies to learning the weights of a perceptron with general activation function in a teacher-student scenario. In particular, we discuss an all-or-nothing phenomenon for the generalization error with a sublinear set of training examples.
引用
收藏
页数:12
相关论文
共 50 条
  • [41] Compressed Sensing With Prior Information: Information-Theoretic Limits and Practical Decoders
    Scarlett, Jonathan
    Evans, Jamie S.
    Dey, Subhrakanti
    IEEE TRANSACTIONS ON SIGNAL PROCESSING, 2013, 61 (02) : 427 - 439
  • [42] An information theoretic rule for sample size adaptation in particle filtering
    Lanz, Oswald
    14TH INTERNATIONAL CONFERENCE ON IMAGE ANALYSIS AND PROCESSING, PROCEEDINGS, 2007, : 317 - +
  • [43] An information-theoretic approach to quantitative association rule mining
    Ke, Yiping
    Cheng, James
    Ng, Wilfred
    KNOWLEDGE AND INFORMATION SYSTEMS, 2008, 16 (02) : 213 - 244
  • [44] An information-theoretic approach to quantitative association rule mining
    Yiping Ke
    James Cheng
    Wilfred Ng
    Knowledge and Information Systems, 2008, 16 : 213 - 244
  • [45] Parallel Rule Induction with Information Theoretic Pre-Pruning
    Stahl, Frederic
    Bramer, Max
    Adda, Mo
    RESEARCH AND DEVELOPMENT IN INTELLIGENT SYSTEMS XXVI: INCORPORATING APPLICATIONS AND INNOVATIONS IN INTELLIGENT SYSTEMS XVII, 2010, : 151 - 164
  • [46] Information theoretic limits to the capacity of volume holographic optical memory
    Neifeld, MA
    Chou, WC
    APPLIED OPTICS, 1997, 36 (02): : 514 - 517
  • [47] Information-Theoretic Limits on the Performance of Auditory Attention Decoders
    Abeysekara, Ruwanthi
    Smalt, Christopher J.
    Karunathilake, I. M. Dushyanthi
    Simon, Jonathan Z.
    Babadi, Behtash
    FIFTY-SEVENTH ASILOMAR CONFERENCE ON SIGNALS, SYSTEMS & COMPUTERS, IEEECONF, 2023, : 1479 - 1483
  • [48] The Capacity of Wireless Networks: Information-Theoretic and Physical Limits
    Franceschetti, Massimo
    Migliore, Marco Donald
    Minero, Paolo
    IEEE TRANSACTIONS ON INFORMATION THEORY, 2009, 55 (08) : 3413 - 3424
  • [49] Information Theoretic Limits of Cardinality Estimation: Fisher Meets Shannon
    Pettie, Seth
    Wang, Dingyu
    STOC '21: PROCEEDINGS OF THE 53RD ANNUAL ACM SIGACT SYMPOSIUM ON THEORY OF COMPUTING, 2021, : 556 - 569
  • [50] Information Theoretic Limits of Improved ACO-OFDM Receivers
    Liu, Xiaozhen
    Zhou, Jing
    Huang, Nuo
    Zhang, Wenyi
    2021 IEEE 94TH VEHICULAR TECHNOLOGY CONFERENCE (VTC2021-FALL), 2021,