Digital Signal Processing Reference
In-Depth Information
Ta b l e 8 . 2
Mean value of
J
with exponent
p
=
1
,
2
,
4
,
8 after 50,000 FEs
Va l u e s o f
J
for different exponents
MEPSO
G3 with PCX
DE
IIWO
J
1
60.3923
95.7113
98.5513
42.5227
J
2
9.0005
10.4252
11.9078
6.4371
J
3
0.5039
0.5732
2.9613
0.3786
J
4
0.0058
0.0178
0.2903
0.0034
Ta b l e 8 . 3
Filter coefficients obtained with exponent
p
=
2 after 50,000 FEs
Filter Coefficients
MEPSO
G3 with PCX
DE
IIWO
p
01
0
.
3061
−
0
.
3016
−
0
.
2426
−
0
.
1652
p
02
0
.
9949
2
.
9023
2
.
4827
−
0
.
7623
p
10
0
.
3935
−
0
.
3435
−
0
.
3484
2
.
7677
p
11
−
0
.
0338
−
2
.
0490
−
2
.
0898
2
.
9921
p
12
0
.
6481
0
.
0387
0
.
0323
−
1
.
6530
p
20
1
.
2345
2
.
4932
2
.
4915
−
1
.
1049
p
21
0
.
5030
0
.
1975
0
.
1613
1
.
7266
p
22
0
.
4481
0
.
7493
0
.
7563
1
.
4750
q
1
−
1
.
0239
−
0
.
4738
−
0
.
9113
−
0
.
3247
q
2
0
.
0342
−
0
.
0843
−
0
.
0255
−
0
.
3656
−
−
r
1
0
.
9605
2
.
9493
2
.
9613
0
.
3232
−
−
−
−
r
2
0
.
0371
0
.
0376
0
.
0344
0
.
1629
s
1
0
.
9523
0
.
8874
0
.
8674
−
0
.
2719
s
2
−
0
.
9056
−
0
.
8476
−
0
.
8075
−
0
.
3672
H
0
0
.
00034
0
.
0784
0
.
0012
−
0
.
0017
8.7 Conclusions
The chapter proposes a new method to the design problem of a zero-phase IIR digital
filter. The proposed algorithm uses a synergy of Temporal Difference Q-Learning
and Invasive Weed Optimization to realize an Adaptive Memetic Algorithm (AMA)
that statistically outperformed the most recent and popular methods outlined in the
existing literature [
4
,
8
,
16
,
18
] in terms of performance accuracy and solution qual-
ity. Further, the superior quality of the recovered image as compared to the other
competitive algorithms demonstrates how the designed 2D filter lends itself to well-
known 2D IIR filter applications like image denoising.
Integration of such reinforcement learning schemes in the evolutionary platform
is a completely new field of research [
1
,
24
,
25
], and further studies will involve
comparative analysis of performance of the proposed memetic algorithm using other
popular reinforcement learning schemes.
Search WWH ::
Custom Search