, ACO algorithms have been tested on a large number of
academic problems. These include problems related to the traveling salesman, as well
as assignment, scheduling, subset, and constraint satisfaction problems. For many of
these, world-class performance has been achieved. For example, ACO algorithms
are, at the time of writing, state-of-the-art (i.e., their performance is comparable to,
or better than, that of the best existing methods other than ACO) for the sequential
ordering problem (Gambardella & Dorigo, 2000), the vehicle routing problem with
time window constraints (Gambardella et al., 1999), the quadratic assignment problem
(Maniezzo, 1999; Stu¨ tzle & Hoos, 2000), the group shop scheduling problem
(Blum, 2003a), the arc-weighted l-cardinality tree problem (Blum & Blesa, 2003), and
the shortest common supersequence problem (Michel & Middendorf, 1999). Additionally,
very good performance has been obtained by AntNet (see chapter 6) on
network routing problems (Di Caro & Dorigo, 1998c).
This success with academic problems has raised the attention of a number of
companies that have started to use ACO algorithms for real-world applications.
Among the first to exploit algorithms based on the ACO metaheuristic is EuroBios
(www.eurobios.com). They have applied ACO to a number of di¤erent scheduling
DITRWSL D13125710
Assignment 2
Paraphrase:
The words highlighted in blue contains a paraphrase of (Huaxin and Shuai, 2011)Link to Article.
The
manipulation of images in the spectrum analyser software will be implemented
by the structural decorator pattern for adding various plots to the
image without inheriting from the image class.”
Inheriting a border from another class puts a border around every
subclass instance” (Gamma,
1995 Page 175). This is clearly what we do not want to do as an image class may have ten or more decorations classes applied, whereas another image class instance could only have one decoration. Fundamentally
the decorator pattern dynamically encloses the image class adding or removing additional
functionality at runtime without
changing the overall structure of the class. (Huaxin and Shuai, 2011) Although can also be error prone as you can add the same functionality more than once. Therefore the decorator pattern must be implemented carefully.
Original Source (Huaxin and Shuai, 2011)
We can use one or more decorators to wrap an object. Given that the decorator has the same super type as the object it decorates, we can pass around a decorated object in place of the original object. The decorator adds its own behaviour either before and/or after delegating to the object it decorates to do the rest of the job. Objects can be decorated at any time, so we can decorate objects dynamically at runtime with as many decorators as we like.
Quotation:
The words highlighted in blue contain a quotation of (Prechelt et al., 2001)
The
importance of these design patterns are recognised by many
researchers. “it is useful to design programs with
design patterns even if the actual design problem is simpler than
that solved by the design patterns”(Prechelt et al., 2001). The nature of the spectrum analyser software and how it handles images makes
this pattern ideal. Therefore the decorator pattern provides a
alternative solution to inheritance especially in the case of our
program as we require to add multiple plots of images at run-time,
wavelet transform, principal components, histogram, scattered plot,
fast Fourier transformation etc.
The
abstract window factory will be implemented for determining what kind
of operating system we need to display the image and plots
essentially decoupling this software from the concrete operating
system at run-time. Most operating systems contain frameworks for
image manipulation......
Summary of a Article: (He et al., 2007)
The words highlighted in blue contains a part summary of (He et al., 2007)
Link
to Article.
Spectroscopy, a
physics theory and the basis was first discovered by Sir Isaac Newton
with his famous splitting of white light via a glass prism to show
primary colours in 1666 Isaac Newton proved that the prism did not
produce the colours but was the light itself. Every element or
molecule emits or absorbs a unique spectra . The spectra is a measure
of wavelengths in the electromagnetic spectrum mostly in the visible
to infra-red light (Born, 1999) Spectroscopy can be used to ascertain
the elements in distance stars Fraunhofer
lines.
Remote sensing of earth from satellites in space. NASA landsat1
satellite was first system of producing multi-spectral data in
digital form.
However
in industry spectrometry is generally a complex procedure that need
to be carried out in a Lab requiring exact measurements. In the
early 2000s with the advancement of ingenious smart phones and
Artificial Intelligence as applied to face recognition and voice (
Hsu et al., 2002) . This research intends at joining Artificial
Intelligence with Spectrometry.
The
main problem with using a neural network as applied to spectroscopy
is basically the noise of the spectrum image and the complexity. A
small spectrum image can be 640 x 480 pixels and in true colour image
this complexity is just too big for a neural net to learn.
(He
et al., 2007) addresses the issue by using a wavelet transform to
reduce the size of the image and as a byproduct removes noise at the same time.
States that such an spectrum image was reduced to 77 wavelet
transforms and that this can reconstruct 99.97% of the original
image. These wavelets are then described by eight principal
components. Making the overall image very easy for the neural network
to process. The principal components are used as inputs and adjusting
the neural network to having 12 neurons in the the hidden level ,
three as output. Claiming a 100% recognition rate of 40 unknown
samples of tea.
References:
Born, M., 1999. Principles of optics: electromagnetic theory of
propagation, interference and diffraction of light, 7th expanded ed.
ed. Cambridge University Press, Cambridge ; New York.
Gamma, E., 1995. Design patterns: elements of reusable
object-oriented software. Addison-Wesley, Reading, Mass.
He, Y., Li, X., Deng, X., 2007. Discrimination of varieties of tea
using near infrared spectroscopy by principal component analysis and
BP model. J. Food Eng. 79, 1238–1242.
Hsu, R.-L., Abdel-Mottaleb, M., Jain, A.K., 2002. Face detection in
color images. IEEE Trans. Pattern Anal. Mach. Intell. 24, 696–706.
Huaxin, M., Shuai, J., 2011a. Design patterns in software
development, in: 2011 IEEE 2nd International Conference on Software
Engineering and Service Science (ICSESS). Presented at the 2011 IEEE
2nd International Conference on Software Engineering and Service
Science (ICSESS), pp. 322–325.
Kay, S.M., Marple, S.L., J., 1981. Spectrum analysis #8212;A modern
perspective. Proc. IEEE 69, 1380–1419.
Prechelt, L., Unger, B., Tichy, W.F., Brossler, P., Votta, L.G.,
2001. A controlled experiment in maintenance: comparing design
patterns to simpler solutions. IEEE Trans. Softw. Eng. 27, 1134–1144.
Final Summary:
Journal of Computer and System Sciences
Seen some very interesting articles in this related to machine learning has lots of other computer resources as well. Some articles are closed but can access.
Relevant to my topics.
Link to Journal:
Relevant to my topics.
Link to Journal:
ISSN:0022
Impact factor 1.0000
ICSE, the International Conference on Software Engineering,
A long running Software engineering Conference (ICSE), and one of the largest , high impact factor, and covers all aspects of software engineering.
Links to Archive:
To main site:
Next Conference: ICSE '14:
36th International Conference on Software Engineering
June 01 - 07, 2014 Hyderabad, India
Impact Factor: 2.05
Links to Archive:
To main site:
Next Conference: ICSE '14:
36th International Conference on Software Engineering
June 01 - 07, 2014 Hyderabad, India
Impact Factor: 2.05
Dr. Dobb's journal: software tools for the professional programmer
Not really a journal as such more a magazine on all aspects of software development, all languages platforms and tools. Back issues in the Kevin street library. The back issues had a high impact factor so good for historic information and programming techniques.
Link To Magazine:
ISSN: 1044-789X
Impact factor: 0.024
Link To Magazine:
ISSN: 1044-789X
Impact factor: 0.024
Subscribe to:
Posts (Atom)