.........Welcome
.........Introduction
.........What is IEC?
.........Research
.........Bibliography
.........Links



State of the art

The research work about IEC algorithms conducted during 1990s was originated by the pioneer contribution to this field made by Dawkins in 1986, [Dawkins 1986].

Inside the IEC field, two different research streams are distinguished:

  • Creative stream: the birth of this stream comes from the research work done with artificial life (AL). AL is focused on the emulation of biological life models through complex algorithms. For this task it is necessary good graphics creation, good image treatment, and quality sounds and music, [Sims 1991b], [Sims 1991c], [Sims 1994], [Dawkins 1986], [Disz 1997], [Unemi 2000] and [Unemi 2003].
  • Humanized technology stream: comes from the research of interaction between humans and computer systems. This stream is more focused on the IEC algorithms interface, trying to decrease the users' fatigue and improving the algorithms' inputs and outputs, making them more efficient. IEC has been expanded to practical fields such as engineering, education, etc, [Parmee 1993], [Ventrella 1994a], [Takagi 1996], [Poli 1997], [Parmee 1998] and [Takagi 1998], unlike the early stages of IEC that was only focused on artistic applications.

Computer graphics (CG)

The first IEC research was the Biomorph of Dawkins, whose polemic theory of evolution, the selfish gene, was useful and motivating for many of the following works, [Ochoa 1998], [Mccormack 1993], and [Smith 2003].

Dawkins used a conventional recursive algorithm with the aim of building trees with an L-system (Lindenmayer). The system called Blind Watchmaker creates evolutionary species called biomorphs that iterate with two operators, which are subjective selection of the L-system outputs and mutation of genes that express the number and angles of L-systems branches, and created several
insect like 2-D CG forms. During this experiment the user acts like a natural selection filter. After several selections of those more appropriate for the user point of view, he or she gets one (more or less) near to the target.

Also there is a large number of applications in this field based on mathematical equations (fractal generation), [Sims 1991a] and [Sims 1992], [Baluja 1993] and [Baluja 1994], [Lund 1995], or [Angeline 1996], [Raynal 1999] and [Lutton 2003], for 3D rendering, [Todd 1991], [Broughton 1997], [Das 1994] and [Tam 2002], for virtual creatures generation, [Sims 1994], [Rowland 2000], or aerodynamic surface design (wings), [NGuyen 1993], [NGuyen 1994] and [NGuyen 1997].

And artistic applications of IEC like Mutator [Todd 1991], [Todd 1994] and [Todd 1999] or [Bentley 1999a], that is used for animation design, cartoon faces design or even screen saver design applications.

Inside the genetic programming (GP) applications environment there is a branch called Interactive Genetic Programming (IGP) with many works done for 3D graphics generation used with a Computer-Animated Virtual Environment, CAVE [Das 1994], [Papka 1996] and [Disz 1997], applications for artistic design or animations based on mathematic equations generated by IGP, [Sims 1991], [Sims 1991a], [Sims 1992], [Sims 1993] and [Min 2004]. Results of this work are Panspermia or Primordial Dance.


Panspermia


Primordial Dance, design with GP

As a good example of IEC and artistic creations is good to remark the project Galapagos, [Sims 1997], and SBART, [Unemi 2000]. Galapagos' project is an IEC application based on L-Systems that allow the visitors to create interesting images that are nowadays exposed in Tokio Multimedia Museum, (NTT Intercommunication Center).

Galapagos: Tokio Multimedia Museum


Galapagos' output sample

SBART is an open source tool for generating graphics with IGP [Unemi 2000]. It uses trees of equations with Genetic Programming (GP), and calculate the value of each pixel. The equations come from the combination of the basic arithmetic operators: exponential, root, sine, cosine, logarithm, maximum and minimum.

The SBART workings is very simple, the user evaluates 20 2D images and after selecting one, the next 20 are generated. This process is repeated until a valid solution is found. Many works can be found in cartoon animation design by computer field. One of the most remarkable is the AnimationLab that allows
to design the motion of one figure interacting with the user. Each part of the figure has an angle and a direction for each extremity, those figures that are capable to walk or run have more chances to be chosen again.


Animation Lab


More detailed information in this field can be found in other works like [McKenna 1990], [Ventrella 1994a], [Ventrella 1994b], or [Ventrella 1995], [Lim 1999] and [Lim 2000].

The Interactive Evolutionary Programming (IEP) has several artistic applications like the fractal generation made by [Angeline 1996], where the user evaluates 10 animations and the system is responsible for the evolution of the selected ones.

Music and sound

IEC was applied to music generation, starting with the first application called GENJAM, [Biles 1994], [Biles 1996] or [Biles 1999] and [Biles 2000].

Another interesting applications are Sonomorph, [Nelson 1993] and [Nelson 1995], or SBEAT, [Unemi 2003], [Horowitz 1994], [Onisawa 2000], [Tokui 2000] and [Fels 2002]. As a result of these applications there are some music songs published that can be heard sometimes in the radio station WDYN. (100.1, New York, USA, WEBPage: http://www.wdyn.net/).

The most common way of working of this type of music generation systems is through the application of specific IEC algorithms, as can be seen in the next figure:


GENJAM squeme

During the evaluation cycle it is important to see how the user interacts with the system, and how the evaluation of music is complex because it must be done sequentially, and with short pieces. GENJAM works with only one masterpiece that evolves in real time while the user evaluates the generated melodies.

Sound and Signal Treatment

The sound and signal treatment is also an interesting field for IEC algorithms. There are applications for speech synthesis, noise reduction (Finite Impulse Response, FIR), [Watanabe 1995] and [Todoroki 2000], hearing aid fitting or sound/signal filtering like the signal treatment as images, [Hsu 2000].

The next figure shows how is the most common scheme for this field.


Scheme for sound and signal treatment with IEC


The work related with speech synthesis can result in different effects in the speaker voice like peace, happiness or fear, based on the phonetic information and the prosodic parameters. Speech has phonetic and prosodic information, and voice impression and naturalness are controlled by prosodic parameters of pitch, amplitude, duration, and speed. The IEC was used to modify prosodic parameters and change voice impressions, [Sato 1996] and [Sato 1997].

The result of the subjective test showed that the quality of the recovered speech was significantly better than original speech, and not only for IGA users, also for other subjects as well. For this purpose the user interaction is absolutely recommendable because the naturalness of the speech can be only determined by another human.

For hearing aid fitting, the combination of IEC techniques to digital signal processing have spread and solved several problems that conventional analog type of hearing aids were unable to overcome. The essential reason to use IEC is that the user is the only one that can identify which combination of filters are more suitable for him, because no one can perceive how another person hears. For this reason the IEC is an interesting and powerful tool to adjust the signal processor parameters [Ohsaki 1998], [Takagi 2000b]. In fact, nowadays there is an application running successfully with real users and excellent results, [Fujii 2000], and recent studies found more satisfaction with IEC systems rather than the traditional ones, [Takagi 1999b], [Fujii 2000] and [Ohsaki 2000].

Image processing

The digital cameras, scanners and PCs, bring to home the latest technology in image capture. Therefore, the digital image processing requirements are growing each day and even more in domestic environments, where the amateur user tries to make up the images with special effects or filters. These tasks that were reserved only for profesional purposes years ago, now is extended for all users.The IEC fits perfectly in this field, because it allows the simple application of different combinations of image effects and filters, and the user only has to evaluate the outputs, choosing those more appropriate for his needs until reaching a good enough solution. The key factor of this scheme is that the user doesn't need to know anything about the complex operations of the image
filtering process.

Another interesting field for image processing is for medical purposes. For example, the IEC can help doctors to enhance medical images, making easy the identification of diseases, as can be seen in the following figure, showing a Magnetic Resonance Image (MRI), [Poli 1997].


MRI enhanced

More interesting works can be found for plant grow monitoring with botanical purposes, [Otobe 1998] and [Tanaka 1999], measuring the image treatment, the growing factor, the leaf colours, shape, and so on.

IEC Related Work

Computer Graphics (CG)
Music & Sound

Sound & Signal Treatment
Image processing
Document design/edition
Industrial Design
E-Commerce & Internet

BDD Searching tools
Data Mining and Knowledge Management
Face Recognition
Robotics
Others

 

2005, University CARLOS III of Madrid