Connecting weighted automata, tensor networks and recurrent neural networks through spectral learning

被引:1
作者
Li, Tianyu [1 ]
Precup, Doina [2 ]
Rabusseau, Guillaume [3 ,4 ]
机构
[1] McGill Univ, Mila, Montreal, PQ, Canada
[2] McGill Univ, CCAI Chair Mila, Montreal, PQ, Canada
[3] Univ Montreal, CCAI Chair Mila, Montreal, PQ, Canada
[4] Univ Montreal, DIRO, Montreal, PQ, Canada
关键词
Weighted automata; Spectral learning; Recurrent neural networks; Tensor networks; Tensor train decomposition; FINITE-STATE AUTOMATA; MATRIX RECOVERY;
D O I
10.1007/s10994-022-06164-1
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 present connections between three models used in different research fields: weighted finite automata (WFA) from formal languages and linguistics, recurrent neural networks used in machine learning, and tensor networks which encompasses a set of optimization techniques for high-order tensors used in quantum physics and numerical analysis. We first present an intrinsic relation between WFA and the tensor train decomposition, a particular form of tensor network. This relation allows us to exhibit a novel low rank structure of the Hankel matrix of a function computed by a WFA and to design an efficient spectral learning algorithm leveraging this structure to scale the algorithm up to very large Hankel matrices. We then unravel a fundamental connection between WFA and second-order recurrent neural networks (2-RNN): in the case of sequences of discrete symbols, WFA and 2-RNN with linear activation functions are expressively equivalent. Leveraging this equivalence result combined with the classical spectral learning algorithm for weighted automata, we introduce the first provable learning algorithm for linear 2-RNN defined over sequences of continuous input vectors. This algorithm relies on estimating low rank sub-blocks of the Hankel tensor, from which the parameters of a linear 2-RNN can be provably recovered. The performances of the proposed learning algorithm are assessed in a simulation study on both synthetic and real-world data.
引用
收藏
页码:2619 / 2653
页数:35
相关论文
共 50 条
  • [41] Learning deep hierarchical and temporal recurrent neural networks with residual learning
    Zia, Tehseen
    Abbas, Assad
    Habib, Usman
    Khan, Muhammad Sajid
    [J]. INTERNATIONAL JOURNAL OF MACHINE LEARNING AND CYBERNETICS, 2020, 11 (04) : 873 - 882
  • [42] Learning deep hierarchical and temporal recurrent neural networks with residual learning
    Tehseen Zia
    Assad Abbas
    Usman Habib
    Muhammad Sajid Khan
    [J]. International Journal of Machine Learning and Cybernetics, 2020, 11 : 873 - 882
  • [43] Riemannian metrics for neural networks II: recurrent networks and learning symbolic data sequences
    Ollivier, Yann
    [J]. INFORMATION AND INFERENCE-A JOURNAL OF THE IMA, 2015, 4 (02) : 154 - 193
  • [44] Mapping additive recurrent neural networks to quantum neural networks
    Rakesh Sengupta
    [J]. Quantum Information Processing, 24 (6)
  • [45] KERNEL LEARNING WITH TENSOR NETWORKS
    Konstantinidis, Kriton
    Li, Shengxi
    Mandic, Danilo P.
    [J]. 2021 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH AND SIGNAL PROCESSING (ICASSP 2021), 2021, : 2920 - 2924
  • [46] Fuzzy finite-state automata can be deterministically encoded into recurrent neural networks
    Omlin, CW
    Thornber, KK
    Giles, CL
    [J]. IEEE TRANSACTIONS ON FUZZY SYSTEMS, 1998, 6 (01) : 76 - 89
  • [47] Towards Interpreting Recurrent Neural Networks through Probabilistic Abstraction
    Dong, Guoliang
    Wang, Jingyi
    Sun, Jun
    Zhang, Yang
    Wang, Xinyu
    Dai, Ting
    Dong, Jin Song
    Wang, Xingen
    [J]. 2020 35TH IEEE/ACM INTERNATIONAL CONFERENCE ON AUTOMATED SOFTWARE ENGINEERING (ASE 2020), 2020, : 499 - 510
  • [48] exploRNN: teaching recurrent neural networks through visual exploration
    Bauerle, Alex
    Albus, Patrick
    Stork, Raphael
    Seufert, Tina
    Ropinski, Timo
    [J]. VISUAL COMPUTER, 2023, 39 (09) : 4323 - 4338
  • [49] exploRNN: teaching recurrent neural networks through visual exploration
    Alex Bäuerle
    Patrick Albus
    Raphael Störk
    Tina Seufert
    Timo Ropinski
    [J]. The Visual Computer, 2023, 39 : 4323 - 4338
  • [50] Structured pruning of recurrent neural networks through neuron selection
    Wen, Liangjian
    Zhang, Xuanyang
    Bai, Haoli
    Xu, Zenglin
    [J]. NEURAL NETWORKS, 2020, 123 : 134 - 141