Neural Network algorithms

New Category!
Thinking of a tool which could fit in this category? Tell us

Total results: 3 nothing-more

What is a neural network?

Neural Networks are computer systems that, using algorithms, attempt to very loosely replicate the model of the human brain on a much smaller scale. These computer systems are able to process information received from external inputs, and can even learn to complete tasks. For instance, neural networks have been developed that can learn to identify certain aspects of images, such as hand-written numbers. From the information that they process, they are able to develop a certain group of characteristics for each number, allowing them to be identified.

How do they work?

These networks are often structured in different tiers of artificial neurons, which each perform different functions in the identification of different things. Signals travel into the network from an external input and through each of the network’s tiers, until they reach the last tier where they are identified. For example, in the identification of hand-written numbers, the 28x28 pixels of the input image are processed by each tier of the network, with each tier identifying certain aspects of the input image. Then, the signals of the input image reach what is called the output layer, where the input image can be identified based on the identification of a series of features of the number’s construction. In a similar way to how children’s brains learn by recognition, these neural networks begin to learn based on the recognition of these input images. These systems often learn by what is called the ‘delta rule’. By this method, neural networks make guesses at what the answer to the input image may be, and determine how they go forward based on how wrong their initial response was.

What are the practical applications of neural networks?

Neural networks give us the ability to find solutions to problems that could otherwise not be solved with conventional computational algorithmic methods. Their utility is enhanced by the fact that they do not need to be programmed because they are self-learning and acquire knowledge through the mistakes they make. For example, these deep layered neural networks can be useful in fingerprint recognition tests, thus giving neural networks a vast array of different uses from the recognition of criminals, to car and bank security processes. Neural networks have also been increasingly utilised in character recognition, with the growth of ‘Palm Devices’ and the use of these kinds of technologies in mobile phones. These networks have also been used as tools to predict stock market fluctuations thanks to their ability to process huge amounts of information extremely quickly. This gives neural networks the opportunity to process all the different factors that could affect stock prices rapidly.

The future of neural networks?

Neural networks, alongside artificial intelligence technologies, look set to continue their rapid growth in the near future. Artificial intelligence is hugely reliant on these brain-replicating networks, allowing artificial intelligence technologies to learn by themselves, without prior programming. A self driving car, for instance, would not require a pre-programmed navigation system, but instead would instead interact with its passenger and understand the desired destination. With the use of neural networks, artificial intelligence technologies would develop the ability to learn and reason in a way which mirrors that of a human. Such technologies may still seem incredibly futuristic, but neural networks will without doubt begin to play an ever increasing role in our daily lives.

keyboard_arrow_downShow more

eloqoon share idea

Have a Neural Network algorithm to share?

Publish a algorithm

Learn more about digital publishing

eloqoon search

Can't find what you are looking for?

Tell us

Or browse through the Catalog

eloqoon search someone

Think of someone who could publish?

Suggest an author

Learn more about digital publishing

Any questions on Neural Networks?

The user community is here to help. Go ahead!

please wait...