Full Analog CMOS Integration of Very Large Time Constants for Synaptic Transfer in Neural Networks

1992 ◽  
pp. 19-33 ◽  
Author(s):  
P. Kinget ◽  
M. Steyaert ◽  
J. Spiegel
Keyword(s):  
Neural Networks ◽  
Large Time ◽  
Time Constants ◽  
Analog Cmos

Pinning observability of competitive neural networks with different time–constants

2019 ◽  
Vol 329 ◽  
pp. 97-102 ◽  
Author(s):  
A. Meyer-Bäse ◽  
A. Moradi Amani ◽  
U. Meyer-Bäse ◽  
S. Foo ◽  
A. Stadlbauer ◽  
...  
Keyword(s):  
Neural Networks ◽  
Time Constants

A Low-Frequency Filter in Inertial Applications

2015 ◽  
Vol 645-646 ◽  
pp. 875-880
Author(s):  
Qiang Li ◽  
Xiao Wei Liu ◽  
Liang Yin ◽  
Jia Jun Zhou
Keyword(s):  
Large Time ◽  
Cutoff Frequency ◽  
Low Frequency ◽  
Frequency Filter ◽  
Time Constants ◽  
Operational Transconductance Amplifiers ◽  
Technical Requirements ◽  
Low Pass ◽  
Transconductance Amplifiers ◽  
Current Shunt

A fourth-order cascade low-pass transconductance capacitor filter is designed, and the coefficients of the filter are calculated. For the implementation of large-time constants the current shunt technique and the current cancellation technique are employed in the operational transconductance amplifiers. The whole filter is simulated in Hspice, the the cutoff frequency is 150.2Hz and the THD of the filter is 65dB, which meet the technical requirements of the inertial applications.

Physically realistic fault models for analog CMOS neural networks

1991 ◽  
Vol 26 (9) ◽  
pp. 1223-1229 ◽  
Author(s):  
D.B.I. Feltham ◽  
W. Maly
Keyword(s):  
Neural Networks ◽  
Fault Models ◽  
Analog Cmos

Learning capacitive weights in analog CMOS neural networks

1994 ◽  
Vol 8 (3) ◽  
pp. 209-225 ◽  
Author(s):  
H. C. Card ◽  
C. R. Schneider ◽  
R. S. Schneider
Keyword(s):  
Neural Networks ◽  
Analog Cmos

Integration and Differentiation in Dynamic Recurrent Neural Networks

1994 ◽  
Vol 6 (3) ◽  
pp. 405-419 ◽  
Author(s):  
Edwin E. Munro ◽  
Larry E. Shupe ◽  
Eberhard E. Fetz
Keyword(s):  
Neural Networks ◽  
Recurrent Neural Networks ◽  
Computational Algorithm ◽  
Network Size ◽  
Sinusoidal Signal ◽  
Dynamic Neural Networks ◽  
Time Constants ◽  
Minimal Network ◽  
Model Networks ◽  
Recurrent Connections

Dynamic neural networks with recurrent connections were trained by backpropagation to generate the differential or the leaky integral of a nonrepeating frequency-modulated sinusoidal signal. The trained networks performed these operations on arbitrary input waveforms. Reducing the network size by deleting ineffective hidden units and combining redundant units, and then retraining the network produced a minimal network that computed the same function and revealed the underlying computational algorithm. Networks could also be trained to compute simultaneously the differential and integral of the input on two outputs; the two operations were performed in distributed overlapping fashion, and the activations of the hidden units were dominated by the integral. Incorporating units with time constants into model networks generally enhanced their performance as integrators and interfered with their ability to differentiate.

Sign in / Sign up

Export Citation Format

Share Document