© The Institution of Electrical Engineers
An analogue MOS circuit that implements a nonlinear Hebbian learning rule is presented. The circuit has two differential inputs (voltages) and yields an output (current) which approximates the product of a cubic and a hyperbolic tangent functions of the two input voltages. This has been successfully incorporated in an analogue integrated circuit implementation of the Herault-Jutten neuromorphic network.
References
-
-
1)
-
C. Jutten ,
J. Herault
.
Blind separation of sources, part I: an adaptive algorithm based on neuromimetic architecture.
Signal Proc.
,
1 -
10
-
2)
-
C. Mead
.
(1989)
, Analog VLSI and neural systems.
-
3)
-
Delbrük, T.: `“Bump” circuits for computing similarity and dissimilarity of analog voltages', Proc. Int. Joint Conf. on Neural Networks, 8th–12th July 1991, Seattle, WA, p. 475–479.
-
4)
-
Palmieri, F., Zhu, J.: `Linear neural networks which minimize the output variance', Proc. Int. Joint Conf. on Neural Networks, 8th–12th July 1991, Seattle, WA, p. 791–797.
-
5)
-
Foldiak, P.: `Adaptive network for optimal linear feature extraction', Proc. Int. Joint Conf. Neural Networks, 18th–22nd June 1989, Washington, DC, p. 401–405.
-
6)
-
Cohen, M.H.: `Analog VLSI implementation of an auto-adaptive synthetic neural network for real-time separation of independent signal sources', 1991, MSE Thesis, The Johns Hopkins University, Baltimore, MD.
-
7)
-
S.J. Orfanidis
.
Gram-Schmidt neural nets.
Neural Computation
,
116 -
126
-
8)
-
Y. Tsividis ,
S. Satyanarayana
.
Analogue circuits for variable-synapse electronic neural networks.
Electron. Lett.
,
1313 -
1314
-
9)
-
E. Vitoz ,
J. Fellrath
.
CMOS analog integrated circuits based on weak inversion operation.
IEEE J. Solid-State Circuits
,
224 -
231
-
10)
-
D.O. Hebb
.
(1949)
, The organization of behaviour.
http://iet.metastore.ingenta.com/content/journals/10.1049/el_19920512
Related content
content/journals/10.1049/el_19920512
pub_keyword,iet_inspecKeyword,pub_concept
6
6