– Electronics manufacturer E.D.&A. uses ‘neural networks’ to make its electronic control systems smarter. (Binary) neural networks are highly effective at recognising patterns and can therefore be used perfectly to tackle problems in a different way.
Prediction vs actual behaviour
Electronics manufacturer E.D.&A. investigated how effectively a neural network can be trained. Take, for example, hob ventilation, where the controls are operated via capacitive touch sensors. These sensors are located beneath a thick glass panel, alongside moisture, dirt and interference from the induction hob itself.
E.D.&A. trained the neural network to recognise the correct signal (e.g. switching on) amidst all this ‘noise’ . The large number of training examples required was obtained using an ‘artificial finger’, which operates the sensor using compressed air, thereby automating the collection of data. The binary variant of neural networks was also further evaluated by E.D.&A. for use in their electronic control systems (microcontrollers). (Binary) neural networks are particularly good at recognising patterns and can therefore be ideally used to tackle problems in a different way.
To illustrate this to machine and equipment manufacturers, E.D.&A. implemented a demo of its own embedded AI (Artificial Intelligence) framework. You draw a digit from 0 to 9 on a touchscreen and it is fed into a binary neural network. This determines which digit you have drawn and displays the result.
What is a ‘neural network’ and how does it work?
An artificial or artificial neural network is both in construction and in operation an imitation of our brain (biological neural network). Among other things, it is used in photo recognition. For example, we can make a system 'smart' so that it can determine whether or not it is a picture of a dog. You obtain a neural network by training it with a long series of examples. Underlying it is long, simple arithmetic that is adjusted based on examples until it gives the correct outcome for all input values. Neural networks are quite strong at recognising patterns. After training, even new pictures can be judged correctly.
Source: press release E.D.&A.
