Page 163 - Electrical Engineering Dictionary

be the percentage of pat- conjunction rule of inference a rule of
tioned. Let e c i ,c j
terns of class c i that are erroneously recog- reasoning which states that if two proposi-
nized as patterns of class c j . The matrix tions A and B are both individually true, then
. n,n
∈ R is called a confusion the combined proposition “A AND B” is also
E = e c i ,c j
matrix. Of course true.
n
X connect/disconnect bus See split trans-
= 1
action.
e c i ,c j
j=1
connected component a maximal-sized
connected region. Also termed “blob.”
congestion a state of a packet-based sys-
tem where too many packets are present in connectedness a graph or subgraph is
the network and the overall performance de- said to be connected if there is at least one
grades. To resolve the congestion, the system path between every pair of its vertices.
must employ some form of congestion con-
trol. See also preventive congestion control connection matrix in a network of gen-
and reactive congestion control. eral topology, the connection matrix identi-
fies how the circuit elements are connected
conical diffraction a scattering phe- together.
nomenon in photorefractive crystals in which
the scattered beam forms a cone of light. connection weight in neural networks,
When a laser beam of finite transverse within the processing element, an adaptive
cross section passes through a photorefrac- coefficient associated with an input connec-
tive crystal, beam fanning often occurs. The tion. It is also referred to as synaptic efficacy.
hologram formed by the incident beam and
the fanned light consists of a multitude of connection-oriented service a mode of
gratings because the fanned light spans a packet switching in which a call is estab-
wide solid angle in space. When such a mul- lished prior to any information exchange tak-
titude of gratings is read out by a laser beam, ing place. This is analogous to an ordinary
only a subset of these gratings matches the phone call, except that no physical resources
Bragg condition with readout beam. The need to be allocated.
wave vectors of the Bragg-matched read-
out beams form hollow cones in momentum connectionist model one of many names
space. Therefore, conical diffractions are ob- given to the learning systems. The notion
served most of the time when fanning occurs. of learning systems has been developed in
Conical diffraction is also often referred to as the fields of artificial intelligence, cyber-
conical scatterings. netics and biology. In its most ambitious
form learning systems attempt to describe or
conicalscattering Seeconicaldiffraction. mimic human learning ability. Attainment
of this goal is still far away. The learning
systems that have actually been implemented
conjugate symmetric transform a prop- are simple systems that have strong relations
erty of a real-valued function that relates to to adaptive control. The learning systems
its Fourier transform. If x(t) is a real-valued are also known under the names of neural
function and its Fourier transform has the nets, parallel distributed processing models,
∗
∗
property that X(−w) = X (w), where de- etc. Examplesoflearningsystemsmostcom-
notes the complex conjugate. The transform monly used are perceptron, Boltzmann ma-
X(w) is said to be conjugate symmetric. chine, Hopfield network. An interesting fea-

c
2000 by CRC Press LLC