A fully-differential CMOS implementation of Oja's learning rule in a dual-synapse neuron for extracting principal components for face recognition
- Spencer, R.G.; Sanchez-Sinencio, E.
Editor(s): Ramirez-Angulo, J.
Analog & Mixed-Signal Center, Texas A&M; Univ., College Station, TX, USA
This paper appears in: Circuits and Systems, 2000. 42nd Midwest Symposium on
On page(s): 1102 - 1104 vol. 2
8-11 Aug. 1999
Las Cruces, NM, USA
2000
Volume: 2
ISBN: 0-7803-5491-5
IEEE Catalog Number: 99CH36356
Number of Pages: 2 vol. (xl+1150)
References Cited: 5
INSPEC Accession Number: 6795188

---

Abstract:
A fully-differential, CMOS implementation of a self-organizing, dual-synapse neuron with on-chip learning for real-time facial feature extraction is presented. The adaptation of the network follows Oja's learning rule and the synaptic weight vector is shown to adapt to the principal component vector of the set of two-dimensional input vectors.

---

Index Terms:
feature extraction; face recognition; neural chips; CMOS analogue integrated circuits; learning (artificial intelligence); SPICE; principal component analysis; self-organising feature maps; operational amplifiers; analogue multipliers; signal flow graphs; fully-differential CMOS implementation; Oja's learning rule; self-organizing dual-synapse neuron; on-chip learning; real-time facial feature extraction; face recognition; synaptic weight vector; two-dimensional input vectors; principal components extraction; eigenvectors; autocorrelation matrix; signal flow graphs; Gilbert cell multiplier; OTA-C integrator; HSPICE