Optical Implementation of a Self-Organizing 
Feature Extractor 
Dana Z. Anderson*, Claus Benkert, Verena Hebler, Ju-Seog Jang, 
Don Montgomery, and Mark Saffman. 
Joint Institute for Laboratory Astrophysics, University of Colorado and the 
Department of Physics, University of Colorado, Boulder Colorado 80309-0.!9 
Abstract 
We demonstrate a self-organizing system based on photorefrac- 
tive ring oscillators. We employ the system in two ways that 
can both be thought of as feature extractors; one acts on a set 
of images exposed repeatedly to the system strictly as a linear 
feature extractor, and the other serves as a signal demultiplex- 
er for fiber optic communications. Both systems implement 
unsupervised competitive learning embedded within the mode 
interaction dynamics between the modes of a set of ring oscilla- 
tors. After a training period, the modes of the rings become as- 
sociated with the different image features or carrier 
frequencies within the incoming data stream. 
1 Introduction 
Self-organizing networks (Kohonen, Hertz, Domany) discover features or qual- 
ities about their input environment on their own; they learn without a teacher 
making explicit what is to be learned. This property reminds us of several 
ubiquitous behaviors we see in the physical and natural sciences such as pat- 
tern formation, morphogenesis and phase transitions (Domany). While in the 
natural case one is usually satisfied simply to analyze and understand the be- 
havior of a self-organizing system, we usually have a specific function in mind 
that we wish a neural network to perform. That is, in the network case we wish 
to synthesize a system that will perform the desired function. Self-organizing 
principles are particularly valuable when one does not know ahead of time ex- 
actly what to expect from the input to be processed and when it is some prop- 
erty of the input itself that is of interest. For example, one may wish to 
determine some quality about the input statistics - this one can often do by ap- 
plying self-organization principles. However, when one wishes to attribute 
some meaning to the data, self-organization principles are probably poor can- 
didates for this task. 
821 
822 Anderson, Benkert, Hebler, Jang, Montgomery, and Saffman 
It is the behavioral similarity between self-organizing network models and 
physical systems that has lead us to investigate the possibility of implementing 
a self-organizing network function by designing the dynamics for a set of opti- 
cal oscillators. Modes of sets of oscillators undergo competition (Anderson, 
Benkert) much like that employed in competitive learning network models. 
Using photorefractive elements, we have tailored the dynamics of the mode in- 
teraction to perform a learning task. A physical optical implementation of self- 
organized learning serves two functions. Unlike a computer simulation, the 
physical system must obey certain physical laws just like a biological system 
does. We have in mind the consequences of energy conservation, finite gain 
and the effects of noise. Therefore, we might expect to learn something about 
general principles applicable to biological systems from our physical versions. 
Second, there are some applications where an optical system serves as an idea] 
"front end' to signal processing. 
Here we take a commonly used supervised approach for extracting features 
from a stream of images and demonstrate how this task can be done in a self- 
organizing manner. The conventional approach employs a holographic corre- 
lator (Vander Lugt). In this technique, various patterns are chosen for recog- 
nition by the optical system and then recorded in holographic media using 
angle-encoded reference beams. When a specific pattern is presented to the ho- 
lographic correlator, the output is determined by the correlation between the 
presented pattern and the patterns that have been recorded as holograms dur- 
ing the 'learning phase'. The angles and intensities of the reconstructed refer- 
ence beams identify the features present in the pattern. Because the 
processing time-scale in holographic systems is determined by the time neces- 
sary for light to scatter off of the holographic grating, the optical correlation 
takes place virtually instantaneously. It is the speed of this correlation that 
makes the holographic approach so interesting. 
While its speed is an asset, the holographic correlator approach to feature ex- 
traction from images is a supervised approach to the problem: an external su- 
pervisor must choose the relevant image features to store in the correlator 
holograms. Moreover the supervisor must provide an angle-encoded reference 
beam for each stored feature. For many applications, it is desirable to have an 
adaptive system that has the innate capacity to discover, in an unsupervised 
fashion, the underlying structure within the input data. 
A photorefractive ring resonator circuit that learns to extract spatially orthog- 
onal features from images is illustrated schematically in figure 1. The resona- 
tor rings in figure i are constructed physically from optical fibers cables. The 
resonator is self-starting and is pumped by images containing the input data 
(White). The resonator learns to associate each feature in the input data set 
with one and only one of the available resonator rings. In other words, when 
the proper feature is present in the input data, the resonator ring with which 
it has become associated will light up. When this feature is absent from the 
input data, the corresponding resonator ring will be dark. 
The self-organizing capabilities of this system arise from the nonlinear dynam- 
Optical Implementation of a Self-Organizing Feature Extractor 823 
 ,_s Acourto-optic 
M odulxa or  
Reflexive 
_rl.l'l. J1Jl  Gain 
ml+ml 
1oo 
multimode fiber 
lllliill 
Figure l: Schematic diagram of the self-organizing photorefractive ring 
resonator. The two signal frequencies, m 1 amd m 2, are identical when 
the circuit is used as a feature extractor and are separated by 280 MHz 
when the system is used as a frequency demultiplexer. 
ics of competition between resonator modes for optical energy within the com- 
mon photorefractive pump crystal (Benkert). We have used this system to ac- 
complish two optical signal processing tasks. In the first case, the resonator 
can learn to distinguish between two spatially orthogonal input images that 
are impressed on the common pump beam in a piece-wise constant fashion. In 
the second case, frequency demultiplexing of a composite input image con- 
structed from two spatially orthogonal image components of different optical 
frequencies can be accomplished (Saffman, 1991b). In both cases, the optical 
system has no a priori knowledge of the input data and self-discovers the im- 
portant structural elements. 
2 A Self-Orgonlzing Photorefractive Ring Resonator 
The experimental design that realizes an optical self-organizing feature extrac- 
tor is shown in figure 1. The optical system consists of a two ring, multimode, 
unidirectional photorefractive ring resonator in which the rings are spatially 
distinct. The resonator rings are defined by loops of 100  core multimode op- 
tical fiber. The gain for both modes is provided by a common BaTiO3 crystal 
that is pumped by optical images presented as speckle patterns from a single 
100  multimode optical fiber. The light source is a single frequency argon-ion 
laser operating at 514.5 nm. The second BaTiO 3 crystal provides reflexive cou- 
pling within the resonator, which ensures that each resonator ring becomes as- 
sociated with only one input feature. 
The input images are generated by splitting the source beam and passing it 
through two acousto-optic modulator cells. The optical signals generated by 
the acousto-optic modulators are then focused into a single 1.5 meter long step- 
index, 100  core, multimode optical fiber. The difference in the angle of inci- 
dence for the two signal beams at the fiber end face is sufficient to ensure that 
the corresponding speckle pattern images are spatially orthogonal (Saffman, 
824 Anderson, Benkert, Hebler, Jang, Montgomery, and Saffman 
1991a). The acousto-optic cells are used in a conventional fashion to shift the 
optical frequency of the carrier signal, and are also used as shutters to impress 
time modulated information on the input signals. When the resonator is oper- 
ating as a feature extractor, both input signals are carried on the same optical 
frequency, but are presented to the resonator sequentially. The presentation 
cycle time of 500 Hz was chosen to be much smaller than the characteristic 
time constant of the BaTiO3 pump crystal. When operating as a frequency de- 
multiplexer, the acousto-optic modulators shift the optical carrier frequencies 
of the input signals such that they are separated by 280 MHz. The two input 
carrier signals are time modulated and mixed into the optical fiber to form a 
composite image composed of two spatially orthogonal speckle patterns having 
different optical frequencies. This composite image is used as the pump beam 
for the resonator. 
3 Unsupervised Competitive Learning 
Correlations between the optical electric fields in images establish the criterion 
for a measure of similarity between different image features. The best mea- 
sure of these correlations is the inner product between the complex-valued spa- 
tial electric field distribution across the input images, 
When S12 = 0 the images are uncorrelated and we define such images as spa- 
tially orthogonal. When the resonator begins to oscillate, neither resonator 
ringhas any preference for a particular input feature or frequency. The system 
modes have no internal bias (i.e., no a priori knowledge) for the input data. As 
the gain for photorefractive two-beam coupling in the common BaTiO3 pump 
crystal saturates, the two resonator rings begin to compete with each other for 
the available pump energy. This competitive coupling leads to 'winner-takes- 
all' dynamics in the resonator in which each resonator ring becomes associated 
with one or the other spatially orthogonal input images. In other words, the 
rings become labels for each spatially orthogonal feature present in the input 
image set. 
Phenomenologically, the dynamics of this mode competition can be described 
by Lotka-Volterra equations (Benkert, Lotka, Volterra), 
dli,p 
dt = Ii'v ( O:i'p - JSi'pli'p 
j,l 
Where Ii, p is the intensity of the oscillating energy in ring i due to energy 
transferred from the input feature p, (Xi, p is the gain for two-beam coupling be- 
tween ring i and feature p, [li,p is the self-saturation coefficient, and Oi, pi},l are 
the cross-saturation coefficients. The self-organizing dynamics are determined 
by the values of the cross coupling coefficients. Thus the competitive learning 
algorithm that drives the self-organization in this optical system is embedded 
Optical Implementation of a Self-Organizing Feature Extractor 
Photorefractive .:i:i!i!i!::ii! i:: 
Crystal .... ::iii::ii::::'!::' 
825 
resonator 
beam 
Figure 2: Reflexive gain interaction. A fraction, 15, of the incident inten- 
sity is removed from the resonator beam, and then coupled back into 
itself by,photorefractive two beam coupling. This ensures 5Vinner- 
takes-all competitive dynamics between the resonator rings. 
within the nonlinear dynamics of mode competition in the pump crystal. 
Once the system has learned, the spatially orthogonal features in the training 
set are represented as holograms in the BaTi03 pump crystal. These holo- 
grams act as linear projection operators, and any new image constructed from 
features in the training set will be projected in a linear fashion onto the learned 
feature basis set. The relative intensity of light oscillating in each ring corre- 
sponds to the fraction of each learned feature in the new image. Thus, the res- 
onator functions as a feature extractor (Kohonen). 
4 Reflexive Gain 
If each resonator ring was single mode, then competitive dynamics in the com- 
mon pump crystal would be sufficient for feature extraction. However, a mul- 
timode ring system allows stability for certain pathological feature extracting 
states. The multimode character of each resonator ring can permit simulta- 
neous oscillation of two spatially orthogonal modes within a single ring. Osten- 
sibly, the system is performing feature extraction, but this form of output is not 
useful for further processing. These pathological states are excluded by intro- 
ducing reflexive gain into the cavity. 
Any system that twists back upon itself and closes a loop is referred to as re- 
flexive (Hofstadter, pg. 3). A reflexive gain interaction is achieved by removing 
a portion of the oscillating energy from each ring and then coupling it back into 
the same ring by photorefractive two-beam coupling, as illustrated in figure 2. 
The standard equations for photorefractive two-beam coupling (Kukhtarev, 
Hall) can be used to derive an expression for the steady-state transmission, T, 
through the reflexive gain element in terms of the number of spatially orthog- 
onal modes, N, that are oscillating simultaneously within a single ring, 
r.._ 
1 + --6 exp 
1-6 N 
Here, exp(G o) is the small signal gain and 15 is the fraction of light removed 
826 Anderson, Benkert, Hebler, Jang, Montgomery, and Saffman 
Ring 
1 
Ring 
2 
Figure 3: Time evolution of the in- 
tensities within each resonator ring 
due to (01 (I1) and (02(12). After 
about 30 seconds, ttie system has 
learned to demultiplex the two input 
frequencies. Ring i has become as- 
sociated with (0 and Ring 2 has 
become associated with to. The con- 
trast ratio between 11 andI 2 in each 
ring is about 40:1. 
from the resonator. The transmission decreases for N > 1 causing larger cavity 
losses for the case of simultaneous oscillation of spatially orthogonal modes 
within a single ring. Therefore, the system favors 'winner-takes-all' dynamics 
over other pathological feature extracting states. 
5 Experimental Results 
The self-organizing dynamics within the optical circuit require several seconds 
to reach steady state. In the case of frequency demultiplexing, the dynamical 
evolution of the system was observed by detecting the envelopes of the carrier 
modulation, as shown in figure 3. In the case of the feature extractor, transient 
system dynamics were observed by synchronizing the observation with the 
modulation of one feature or the other, as shown in figure 4. The frequency de- 
multiplexing (figure 3) and feature extracting (figure 4) states develop a high 
contrast ratio and are stable for as long as the pump beam is present. Measure- 
ments with a spectrum analyzer show an output contrast ratio of better than 
40:1 in the frequency demultiplexing case. 
The circuit described here extracts spatially orthogonal features while contin- 
"'"7  .... ! ! I. I. 
** : 
0.8 
0,6 
0.4 
O2 
0 
1 
oJ 
o 
0.4 
Figure 4: Time evolution of the 
intensities in each resonator ring 
due to the two input pictures. 
The system requires about 30 sec- 
onds to learn to extract features 
from the input images. Picture 1 
is associated with Ring I and pic- 
ture 2 is associated with Ring2. 
Optical Implementation of a Self-Organizing Feature Extractor 827 
uously adapting to slow variations in the spatial mode superposition due to 
drifts in the carrier frequency or perturbations to the fibers. Thus, the system 
is adaptive as well as unsupervised. 
6 Summary 
An optical implementation of a self-organizing feature extractor that is adap- 
tive has been demonstrated. The circuit exhibits the desirable dynamical prop- 
erty that is often referred to in the parlance of the neural networks as 
'unsupervised learning'. The essential properties of this system arise from the 
nonlinear dynamics of mode competition within the optical ring resonator. The 
learning algorithm is embedded in these dynamics and they contribute to its 
capacity to adapt to slow changes in the input signal. The circuit learns to as- 
sociate different spatially orthogonal images with different rings in an optical 
resonator. The learned feature set can represent orthogonal basis vectors in an 
image or different frequencies in a multiplexed optical signal. Because a wide 
variety of information can be encoded onto the input images presented to the 
feature extractor described here, it has the potential to find general application 
for tasks where the speed and adaptability of self-organizing and all-optical 
processing is desirable. 
Acknowledgements 
We are grateful for the support of both the Office of Naval Research, contract 
#N00014-91-J-1212 and the Air Force Office of Scientific Research, contract 
#AFOSR-90-0198. Mark Saffman would like to acknowledge support provided 
by a U.S. Air Force Office of Scientific Research laboratory graduate fellowship. 
References 
D.Z. Anderson and R. Saxena, Theory of Multimode Operation of a Unidirec- 
tional Ring Oscillator having Photorefractive Gain: Weak Field Limit, J. Opt. 
Soc. Am. B, 4, 164 (1987). 
C. Benkert and D.Z. Anderson, Controlled competitive dynamics in a photore- 
fractive ring oscillator: 'Winner-takes-all" and the "voting-paradox" dynamics, 
Phys. Rev. A, 44, 4633 (1991). 
E. Domany, J.L. van Hemmen and K. Schulten, eds., Models of Neural Net- 
works; Springer-Verlag (1991). 
T.J. Hall, R. Jaura, L.M. Connors and P.D. Foote, The Photorefractive Effect- 
A Review; Prog. Quant. Electr., 10, 77 (1985). 
J. Hertz, A. Krogh and R.G. Palmer, Introduction to the Theory of Neural Com- 
putation; Addison-Wesley (1991). 
D. R. Hofstadter, Metamagical Themas: Questing for the Essence of Mind and 
Pattern; Bantam Books (1985). 
828 
Anderson, Benkert, Hebler, Jang, Montgomery, and Saffman 
T. Kohonen, Self-Organization and Associative Memory, 3rd Edition; Springer- 
Verlag (1989). 
N. V. Kukhtarev, V.B. Markov, S.G. Odulov, M.S. Soskin and V.L. Vinetskii, 
Holographic Storage in Electrooptic Crystals. I. Steady State; Ferroelectrics, 
22, 949 (1979). 
N. V. Kukhtarev, V.B. Markov, S.G. Odulov, M.S. Soskin and V.L. Vinetskii, 
Holographic Storage in Electrooptic Crystals. II. Beam Coupling - Light Ampli- 
fication; Ferroelectrics, 22, 961 (1979). 
A.J. Lotka, Elements of Physical Biology; Baltimore (1925). 
M. Saffman and D.Z. Anderson, Mode multiplexing and holographic demulti- 
plexing communications channels on a multimode fiber, Opt. Lett., 16, 300 
(1991a). 
M. Saffman, C. Benkert and D.Z. Anderson, Self-organizingphotorefractive fre- 
quency demultiplexer, Opt. Lett., 16, 1993 (1991b). 
  Vander Lugt, Signal Detection by Complex Spatial Filtering; IEEE Trans. 
Inform. Theor., IT-10, 139 (1964). 
V. Volterra, Leqons sur la Thorie Mathematiques de la Lutte pour la Vie; 
Gauthier-Villars (1931). 
J.O. White, M. Cronin-Golomb, B. Fischer, and , Yariv, Coherent Oscillation 
by Self-Induced Gratings in the Photorefractive Crystal BaTi03; Appl. Phys. 
Lett., 40, 450 (1982). 
PART XII 
LEARNING AND 
GENERALIZATI ON 
