ORIGINAL ARTICLE
Memristor-based chaotic neural networks for associative memory
Shukai Duan
•
Yi Zhang
•
Xiaofang Hu
•
Lidan Wang
•
Chuandong Li
Received: 3 September 2013 / Accepted: 19 May 2014 / Published online: 6 June 2014
Ó Springer-Verlag London 2014
Abstract In chaotic neural networks, the rich dynamic
behaviors are generated from the contributions of spatio-
temporal summation, continuous output function, and
refractoriness. However, a large number of spatio-temporal
summations in turn make the physical implementation of a
chaotic neural network impractical. This paper proposes
and investigates a memristor-based chaotic neural network
model, which adequately utilizes the memristor with
unique memory ability to realize the spatio-temporal
summations in a simple way. Furthermore, the associative
memory capabilities of the proposed memristor-based
chaotic neural network have been demonstrated by con-
ventional methods, including separation of superimposed
pattern, many-to-many associations, and successive learn-
ing. Thanks to the nanometer scale size and automatic
memory ability of the memristors, the proposed scheme is
expected to greatly simplify the structure of chaotic neural
network and promote the hardware implementation of
chaotic neural networks.
Keywords Memristor Chaotic neural network
Superimposed pattern Many-to-many association
Successive learning
1 Introduction
Chaotic behaviors probably exist in biological neurons. In
particular, chaos is considered to play a crucial role in
associative memory and learning in human brains. Aihara
et al. [1] studied and modeled the chaotic responses of a
biological neuron and proposed the concept of chaotic
neural network (CNN) in 1990. In the past several decades,
chaotic neural networks have been extensively investigated
[2–10]. Many characteristics and advantages of CNNs of
such as high computation efficiency and adaptability have
been explored in a variety of employments, including
associative memory [4–10], pattern recognition [2], and
combinatorial optimization [3]. However, the continuous
development of CNNs has been slowed down for the dif-
ficulty in physical implementation. In other words, the
complexity of the traditional chaotic neural networks leads
to the challenges in hardware circuit implementation and
limited network scale, which in turn restricts its informa-
tion processing capability, thus reducing its practical
applications.
The memristor, called the fourth fundamental element,
may bring new hope to this canonical research field. In
1970s, Leon Chua explored the missing relationship
between flux (u) and charge (q) of a device based on the
symmetry arguments of circuit theory and thus theoreti-
cally formulated the memristor [11] and memristive sys-
tems [12]. About 40 years later, Williams and his team at
the Hewlett-Packard (HP) Labs announced that they
experimentally confirmed the existence of the memristor
and successfully developed an effective electronic device
with nanometer oxides thin film structure [13]. Since the
exciting progress, increasingly much attention from aca-
demic and industry circles have been paid on the potential
element. Gradually, the charming properties of the
S. Duan (&) Y. Zhang L. Wang C. Li
School of Electronics and Information Engineering, Southwest
University, Chongqing 400715, China
e-mail: duansk@swu.edu.cn
X. Hu (&)
Department of Mechanical and Biomedical Engineering, City
University of Hong Kong, 83 Tat Chee Avenue, Hong Kong,
China
e-mail: xiaofanghs@gmail.com
123
Neural Comput & Applic (2014) 25:1437–1445
DOI 10.1007/s00521-014-1633-x