The neural networks of the brain are made up of interlinked neurons. Particular links are reinforced when the brain experiences sensations (touch, smell, vision, etc.). For instance, on detecting the smell of frying bacon, links may be reinforced between the clusters of neural nets which record the components of that type of smell, that type of image (bacon in various stages of cooking), the sound of sizzling, the sensation of heat from the cooker, anticipated taste and so on, forming a complex association.
This concept has been used to create artificial neural nets (ANNs) which, though less complex than those in the brain, can be trained to recognise input patterns. If, for instance, one were trying to train an ANN to recognise a military tank, one could feed various images (including tanks in all their variety and viewpoints) to the net and reinforce the connections when the correct answer is given.
I recall reading about an attempt to train such a net to differentiate hostile and friendly tanks. The experiment seemed to go well in the training using photographs, but when more extensive trials were undertaken it all seemed to go wrong; all the tanks were classed as friendly. It was eventually realised that the training set had consisted of images of friendly tanks in operation in the field and hostile tanks from photographs in books, side views, etc. The neural net was taking the presence of trees as its indicator of friendliness.
Inspec comprehensively covers this wide subject. Papers can be found in Section C (Computer and Control Abstracts: Chapters C12, C5, and C7), Section B (Electrical and Electronics Abstracts: Chapter B12 and B41), and Section A (Physics Abstracts: Chapter A87). The main codes are shown below:
Classification Codes: Neural Networks
Systems theory and cybernetics
C1230D: Neural nets
This subsection is used for theoretical aspects of neural nets
C1290L: Systems theory applications in biology and medicine
Biological neural nets and thought process models can be found here.
Computer hardware
C5190: Neural net devices
This code includes neural chips
C5290: Neural computing techniques
Neural network architectures and applications can be found here
C5340: Associative storage
Neural net storage can be found here
Circuit theory and circuits
B1295: Neural nets (circuit implementation)
Biophysics, medical physics, and biomedical engineering
A8730: Biophysics of neurophysiological processesIn this code and subsections, biological aspects of neural nets can be found.