The human brain is the most perfect information processing system. It has always been a scientist's dream to study the human brain from the perspective of information processing and develop a computer that can "think" like the human brain. In the early 1980s, in the United States, Japan, and then in China, there was an upsurge of studying neural network theory and neural computer.
Many microprocessors are used to imitate the neuron structure of the human brain, and a neural computer is formed by adopting a structural design similar to that of the human brain. In addition to many processors, there are nerve-like nodes, and each node is connected with many other nodes. If every operation is assigned to each microprocessor and they operate at the same time, its information processing speed and intelligence will be greatly improved. Scientists predict that in the future, with supercomputers made by nanotechnology, it is entirely possible to simulate computers with human intelligence. This kind of computer is also called artificial brain.
For German neuroscientist Peter Fremoz, the goal of developing a neural computer is a bit too far. He is studying how to combine biological organisms with silicon chips to study the self-learning and memory of neurons.
Last year, the research team led by Fremau Z fixed two snail neurons in the middle of a silicon chip, which looked like a sharp "fence" etched on the chip and surrounded the neurons. In the next two days, the neurons of the two snails grow synapses and connect with each other. They can also exchange electrical signals with each other or with the electrodes on the chip.
The connection of neurons made Fremoz clearly see the reaction of cells to these electrical signals. With the adoption of more neurons, he plans to study the physical changes and memory storage of neural networks. Fremau said, "We have the most basic components, and we can combine digital electronic components with neural networks. The next step is to make more neurons on the silicon chip. The goal is to create a small self-learning network. " Scientists at Duke University in the United States are developing a "monkey brain computer". They want to understand and develop nerve repair for the paralyzed. At present, his research team is trying to get the monkey's brain to send signals to control the robot's arm. When monkeys reach for food, microelectrodes embedded in their cerebral cortex will read nerve signals. The computer analyzes these signals, distinguishes the pattern of brain activity, and predicts the movement direction of monkey's upper limbs, thus guiding the robot's arm movement. In the experiment, when the monkey moves the upper limb, the robot's arm moves with it, and the movements are surprisingly coordinated.
Scientists conducting this experiment believe that in the future, the human brain may be connected with other external human brains or computers through wires, which can directly transmit signals and receive feedback. This technology can be used to create a virtual reality system. In such a system, astronauts who landed on Mars can learn how to deal with gravity problems on Mars before leaving the Earth.
Russian scientists also studied the imitation of the human brain and developed the first artificial brain in 2R: a "neural computer" with the same wisdom as the human brain.
Valtsev, a Russian scientist, said that Russia's new computer imitated the operation mode of brain cells (or neurons), adopted the latest findings of neurophysiology and neurophysiology, surpassed the past brain model, and created a truly thinking machine. But he warned that this scientific breakthrough also has its potential dangers. He said that if the new artificial brain is not handled properly, it will become a scientific monster. He said: "This machine must be trained like a newborn. It is very important to make it our friend, not a criminal or an enemy. "
Japanese scientists have developed the large-scale integrated circuit chips needed to manufacture neural computers. On the 1.5 square centimeter silicon chip, 40,000 neurons and 40,000 ganglia can be set, and this chip can achieve the operation speed of 200 million times per second. The neural computer developed by Fujitsu Research Institute updates data nearly 1000 billion times per second. NEC has introduced a neural network speech recognition system, which can recognize incoming and outgoing voices, with a correct rate of 99.8%. The United States has developed a neural computer, which consists of two nerve blocks: the left brain and the right brain. The right brain is an empirical function part, with more than 10000 neurons, which is suitable for image recognition. The left brain is a part of the recognition function, which contains 6.5438+0 million neurons and is used to store words and grammar rules. Now, airports in new york, Miami and London have used neural computers to check explosives, and 600 ~ 700 pieces of luggage can be checked every hour, with a detection rate of 95% and an error rate of 2%.
Neurocomputers will be widely used in various fields. It can identify the characters, symbols, graphics, languages and signals received by sonar and radar, explain and inspect, estimate the market, analyze new products, make medical diagnosis, control intelligent robots, realize the automatic driving of cars and planes, find and identify military targets, and make intelligent decisions and orders.