Neural networks at work
Article Abstract:
Neural networks are increasingly utilized in image and voice recognition and in financial forecasting because they can recognize patterns well, are accurate and cost-effective. Neural networks are effective for optical character recognition (OCR) because they can handle imprecise data and complex situations, such as the huge set of characters in written Japanese. Neural networks are effective for magnetic-ink character recognition (MICR) because they are very accurate, economical and can make devices easy to use. Function estimation needs neural networks to accurately represent real-world systems or functions. Financial forecasters teach neural networks to imitate a market and inform their investments. Neural networks are helpful in process control because they can make predictive models based on the large amounts of information most companies collect about processes but without theory, which companies often do not have.
Publication Name: IEEE Spectrum
Subject: Engineering and manufacturing industries
ISSN: 0018-9235
Year: 1993
User Contributions:
Comment about this article or add new information about this topic:
Neural networks made easy
Article Abstract:
'Neural Network Toolbox' is a new software that can be used with the Matlab language to construct artificial neural networks and teach them to students and professionals. Matlab is used to solve complex mathematical problems. The 'Neural Nerwork Toolbox' requires a 386, a 486-based personal computer or a Macintosh system and more than 16 megabytes of memory when used on the network. The package helps several users to use the Matlab software to solve problems simultaneously. The package costs $195 and is highly versatile.
Publication Name: IEEE Spectrum
Subject: Engineering and manufacturing industries
ISSN: 0018-9235
Year: 1995
User Contributions:
Comment about this article or add new information about this topic:
Neural networks automate inspections
Article Abstract:
Neural networks have great potential in doing intelligent product inpection work. Neurocomputers are given a set of data from which they learn the important elements about the data. The lessons learned are used in detecting, identifying, predicting and solving problems. Neurocomputers can be programmed to accept a product if it is good or reject a bad product. It is not required to define what good or bad means because neurocomputers learn what these qualities are based on the images they have studied.
Publication Name: Quality
Subject: Engineering and manufacturing industries
ISSN: 0360-9936
Year: 1993
User Contributions:
Comment about this article or add new information about this topic:
- Abstracts: The network not taken. Spectral lines. Old world, new grid
- Abstracts: Digital Processing and Storage Helps Picture. Neural engineering's image problem. Art Conservators Embrace Digital Image Processing
- Abstracts: Innovative genius: the innovative mind at work. How Israeli engineers improve U.S. weaponry
- Abstracts: Quality water heating on tap. Quality management. The quest for quality
- Abstracts: Learning new skills. Learning curve boomerangs. Selecting managers: how to avoid the Peter Principle