Neural Networks

Neural networks, which are also called artificial neural networks, are an information processing paradigm that imitates biological nervous systems, such as the brain. The biological neural network inside the human brain is composed of a group or groups of chemically connected or functionally associated neurons, which could perform a specific physiological function, such as pattern recognition and decision making. Similarly, the artificial neural network system is composed of many simple processing elements operating in parallel. The function of the processing elements is for example determined by network structure and connection strengths.

First attempt of neural network research goes back to 1943 when the first artificial neuron was produced by the neurophysiologist McCulloch and the logician Walter Pits. However the computing technology was limited at that era, and they couldn’t do anything further; but their studies laid foundations for later work in neural network research. [1]

So far, the neural network has been a real movement of the discipline in three different directions:

Artificial neural networks (ANNs)
Neuromorphic Systems
Computational Neuroscience

Sunday, May 4, 2008

Artificial neural networks (ANNs)

The power and usefulness of artificial neural networks have been demonstrated successfully in several real world applications including speech synthesis, diagnostic problems, biomedicine, business and finance, robotic control, signal processing, computer vision and many other problems that fall under the category of pattern recognitions [2,3].

Example: Artificial Neural Networks(ANNs) in medicine and cancer research

Artificial Neural Networks are currently a popular research field in medicine. At the moment, the research is mostly focused on modeling and recognizing disease or cancers inside the human body. For example, many complex factors are involved in the prediction of several diseases or cancers. Therefore, ANNs could be utilized to evaluate complex nonlinear relations among many single variables and have been demonstrated to be promising tools for improving diagnosis, staging, and prognosis of prostate cancer with the accuracy of 79–84%.[4]

Interview with Prof. Azzam F.G. Taktak in Department of Clinical Engineering at Royal Liverpool University Hospital, UK. His research areas focus on computation and biomedicine, and he has served as the reviewer for various scientific journals, such as Artificial Intelligence in Medicine, International Journal of Artificial Intelligence Tools, Neural Networks, and European Journal of Neurology. He had published several papers and one of those is a review article to discuss the advantages of using ANNs as decision making tools in the field of cancer. [5][Dr. Azzam F.G. Taktak’s professional link]

In the interview I conducted with Prof. Taktak to discuss the development and future of ANNs in biomedicine, he first clarified that many people have some misconception about the ANNs because they think ANNs work like “black magic”. He said that in real case ANNs technologies have solid theoretic foundation and he also emphasized that ANNs and other Machine learning algorithms are very useful tools in analyzing problems that are too complex for the human brain to solve. They have a major role to play in high-dimensional datasets (such as micro array data) where traditional statistical models fail. Nevertheless, the only drawback is that a very large number of samples are required to make sure that the answers they provide is a sensible one. “And such datasets are not yet widely available” he said.

Since the limitation lies with the quality of the data rather than the model, Prof. Taktak is hoping to see more emphasis on real-world applications on the use of AI/ANNs rather than focusing on developing the methodology alone. And he anticipated that ANNs could be of great value to medical application when the accumulated clinical dataset is available in the future.

Other useful websites or links:
Artificial Neural Networks in Medicine World Map
Field distribution of ANNs
Real world application of ANNs
Experts in ANNs
Neural betwork centers worldwide

Posted by at

No comments:

Post a Comment

Subscribe to: Post Comments (Atom)

Conclusion

Neural networks possess the ability to learn from examples, which makes them very powerful, and have extended the world of computing and improved a lot for real life application as well. Neural networks also contribute to areas of neuroscience and neurophysiology, which shed the light on mimicking or synthesizing human brains. Future application of neural networks will have more potential when it is integrated with other AI computing approaches, such as fuzzy logics and robotics, and provide the better environment for human life.

Bibliography

1. Müller, Berndt, Joachim Reinhardt, and Michael Thomas Strickland. Neural Networks: An Introduction. 2nd ed. New York: Springer-Verlag, 2002.

2. Lisboa, Paulo J., Emmanuel C. Ifeachor, and Piotr S. Szczepaniak. Artificial Neural Networks in Biomedicine. London: Springer-Verlag, 2000.

3. Krogh, Anders. “What are Artificial Neural Networks?” Nature Biotechnology 26 (2008): 195-197.

4. Carsten, Stephan, Henning Cammann, and Klaus Jung. “Artificial Neural Nnetworks: Has the Time Come for Their Use in Prostate Cancer Patients?” Nature Clinical Practice. Urology 2 (2005): 262-263.

5. Lisboa, Paulo J., and Azzam F.G. Taktak. “The Use of Artificial Neural Networks in Decision Support in Cancer: A Systematic Review” Neural Networks 19 (2006): 408–415.

6. Silver, Rae, Kwabena Boahen, Sten Grillner, Nancy Kopell, and Kathie L. Olsen. “Neurotech for Nneuroscience: Unifying Concepts, Organizing Principles, and Emerging Tools” Journal of Neuroscience 27 (2007):11807-11819.

7. Boahen, Kwabena. “Brains in Silicon.” Department of Bioengineering at Stanford University. 28 Apr. 2008

8. Lynch, Zack. “Blue Brain Project Moves onto Whole Brain, Really?” Corante.com 28 Nov. 2007. 26 Apr. 2008. 2007/11/28/ blue_brain_project_moves_onto_whole_brain_really.php>

9. MArkram, Henry. “The Blue Brain Project.” Nature Reviews Neuroscience 7 (2006): 153-160.

10. Blue Brain Project. 2008. IBM and EPFL 20 Apr. 2008

11. Bressan, Beatrice. “The Greatest Challenge: Computing the Brain.” Interview. Cerncourier.com 20 Aug. 2007