Blockbuster or Flop?

Neural networks are drastically changing the way we do things and leading us into the future in great strides. There is virtually limitless potential when it comes to this new technology. That being said, neural networks are not perfect, and still have a long way to go before their bright promises become realities. This technology is still in the early stages of development, but its fairly impressive statistical accuracy has found practical use in niches of the marketplace as they await further development.

One very interesting market in which neural networking has proved useful is the movie industry. When a big Hollywood movie producer expresses interest in sinking hundreds of millions dollars into the next hopeful box office smash, he would like to know, or at least have some assurance that the movie will be successful beforehand. Entertainment industry market researcher Edith Bodnar came up with the idea that an artificial neural network can learn the seven key parameters that influence the likelihood of a movie's success, and judge a movie based on these criteria before it ever goes into production. Ramesh Sharda, a computer expert from Oklahoma State University, employed such a neural network, using data on 834 movies released between 1998 and 2002. Each movie was evaluated by this ANN based on

• "Star value" of the cast
• movie's age rating
• time of release against the competition
• film genre
• degree of special effects
• sequel or not
• # of screens expected to open in
  

This information was used to place the movie into one of nine categories from "flop"(< $1M) to "blockbuster"(>$200M). Of the data analyzed, the system reported revenue category of the films dead-on accurate 37% of the time. Although not incredibly accurate, it did report accurately to within one category to the left or right an impressive 75% of the time. While not a perfect science, it is certainly a "powerful decision aid," according to Bodnar, and a bright and promising illustration of where neural networks are headed in the future.






"Neural Network Sorts the Blockbusters from the Flops." 15 December 2005. New Scientist Magazine issue 2530

Neural Networks and Manufacturing

In the manufacturing industry artificial neural networks are used to perform manufacturing functions, to control the quality of the finished product and to maintain machine used to produce it. In machines that do more than one job process neural networks work by telling the matching when one process is complete and how to move on to the next one. It also tells the machine in some cases when and how deep to cut raw materials such as glass. Because these raw materials have unpredictable depths or widths, a neural networks act like a human eye to eliminate waste and try to create uniformity from unalike materials.


To control quality of a product at the end of production, neural networks are used because they are faster and more efficient. In most print shops, printers use sensor that evaluated the amount of ink that is on each paper. If a paper doesn’t meet the preset specifications the whole production line is stopped and an employee must expel the bad printing. Other times the bad printing is expelled through a complicated system which starts when the sensor deems a printing bad, but never stopping production. Neural networks can also sort items by weight and color. The cranberry industry uses this technology to separate cranberries for their end use. Neural networks use channels sending different cranberries through different processes. Smaller cranberries are used for jellied sauce, white cranberries used to make juice, and the bigger redder ones are sold whole as is.


Neural networks are used to measure the fitness of certain machine components. Often in manufacturing blades get dull. Blades used for cutting raw materials are now being equipped with sensors that measure the vibration and tension created by cutting. When the vibration tells the sensor that the blade is weak, the machine either shuts off so the blade can be changed or in some cases changes the blade itself. This curtails the amount of material that would have to be discarded because the cut was not up to par.

Rajagopalan, Ramesh, and Purnima Rajagopalan. "Applications of Neural Network in Manufacturing." Proceedings of the 29th Annual Hawaii International Conference on System Sciences (1996): 447-454.

So What is Neural-Networking Technology??

Real-time data mining, powered by neural-network technology is changing the way people run businesses. Neural networking technology is an artificial intelligence system that is capable of finding and differentiating patterns. The brain learns through experience and that is what the goal of a neural network system is, and it is seen as the next major leap in the computer industry. Even though the technology may seem new to us as consumers, neural networks have actually been around for years, starting in the Pentagon's Defense Advanced Research Projects Agency (DARPA) and a firm founded by technology visionary Robert Hecht-Nielsen. But even though some of the simplest of animals can detect pattern changes, it is something very difficult to accomplish with a computer. But, advance in biological research have opened doors to understanding how the brain stores information as patterns.

So how is this new technology being utilized? "Traditional fraud detection operates with a delay of months or years," Tammy Delatorre, a spokesperson for Fair Isaac, told TechNewsWorld. "While useful, this approach does little to prevent fraudsters from committing costly fraud schemes and then disappearing with the money." What neural-networking technology has done is it has brought fraud detection to the next level. How the technology works is it learns the clients purchasing patterns and scores it. When the pattern is changed, the score changes and the right people are alerted. "The score allows claims professionals to determine which claims must be taken out of the payment stream for further investigation, and allows the rest of the claims to be fast-tracked for payment," said Fair Isaac's Delatorre. "The system also provides reason codes to help investigators determine the most appropriate action on each high-risk claim."

The technology is also being used in the customer service area of business. Customer-behavior prediction is exactly as it sounds, it tries to predict the buying patterns of customers in the future. This is a job that used to take a lot of time and research for companies, but now is being done via computer, not just seasonally, but hourly. The software can show the different things people are buying at different times of the year, month, week, and day! It then predicts how much of each product needs to be ordered in advance to meet these forecasted needs.

Neural-networking technology is making our world different everyday, and if you haven’t noticed, you are just not paying attention. You shop online now and before you finish, the website is already giving you recommendations for future purchases based on past purchases. As this new technology finds new avenues to advance service, it is exciting to see where it will take us.

Koprowski, Gene. "Neural-Network Technology Moves into the Mainstream." 07 June 2003 09 Feb 2008 .

Mitlia v.s. Neural Networks

In today’s battlefield, many obstacles are met in the process of executing mission. Loads of research has been done in an attempt to reduce the amount of hurdles which needed to be overcome. One of significance is fratricide, the military term for being killed by your own side; numerous soldiers have lost there lives in this manner, defending our country. This is a result of platoons’ lack of ability to positively identify friendly forces. Military forces are now applying the latest technology in an effort to prevent these occurrences, by utilizing Neural Network based technology.

This new innovative technology is called Battlefield Target Identification Device. It uses neural networks learning ability to instantaneously indentify battlefield targets. All transponder equipped friendly vehicles can be identified. When a query is made, the friendly vehicle will respond. After various test, this new system has a ninety-eight percent positive identification rate. This enables soldiers on the battlefield to make the right decision in terms of applying deadly forces to the opposition. As a soldier, you can now be very certain that the opposing target you are about to engage is not one of your own. With integrated weapons and almost all of the uncertainty eliminated, mission efficiency is greatly increased. As the system is continuously trained, it will be capable of recognizing individual faces to further reduce the chance of fratricide.

Perhaps the notorious Pat Tillman, an ex-Nation Football League player turned United States soldier would still be alive. He was accidentally engaged by a friendly platoon and executed. He is just one of the many whom have fallen victim to fratricide. However, with the implementation of neural network systems in today’s battlefield, these unfortunate occurrences will hopefully be ceased.