Target response, neural network

Artificial neural networks are now widely used in science. Not only are they able to learn by inspection of data rather than having to be told what to do, but they can construct a suitable relationship between input data and the target responses without any need for a theoretical model with which to work. For example, they are able to assess absorption spectra without knowing about the underlying line shape of a spectral feature, unlike many conventional methods. 

ANNs are the favorite choice as tools to monitor electronic noses,8 where the target response may be less tangible than in other studies (although, of course, it is still necessary to be able to define it). Many applications in which a bank of sensors is controlled by a neural network have been published and as sensors diminish in size and cost, but rise in utility, sensors on a chip with a built-in ANN show considerable promise. Together, QSARs and electronic noses currently represent two of the most productive areas in science for the use of these tools. 

There are problems with this approach since enzymes isolated from natural sources such as the electric organ of electric eels often display low sensitivity and selectivity to the wide range of potential pesticide targets [21]. A possible solution to this is the development of a multisensor array where a variety of genetically modified acetylcholinesterases are immobilised on an array of electrochemical sensors and the responses from these are then processed via a neural network. 

There are two learning paradigms that determine how a network relates to its environment. In supervised learning (learning with teacher), a teacher provides output targets for each input pattern, and corrects the network s errors explicitly. The teacher has knowledge of the environment (in the form of a historical set of input-output data) so that the neural network is provided with desired response when a training vector is available. The... 

Clinical evidence, lesion, and stimulation studies all point toward the participation of vitally important neural sites in the control of saccades, including the cerebellum, superior colliculus (SC), thalamus, cortex, and other nuclei in the brain stem, and that saccades are driven by two parallel neural networks [Enderle, 1994, 2002]. From each eye, the axons of retinal ganglion cells exit and join other neurons to form the optic nerve. The optic nerves from each eye then join at the optic chiasm, where fibers from the nasal half of each retina cross to the opposite side. Axons in the optic tract synapse in the lateral geniculate nucleus (a thalamic relay), and continue to the visual cortex. This portion of the saccade neural network is concerned with the recognition of visual stimuli. Axons in the optic tract also synapse in the SC. This second portion of the saccade neural network is concerned with the location of visual targets and is primarily responsible for goal-directed saccades. 

When dealing with complex matrices such as those encountered in food and medical samples, it is often difficult to develop a sensor that exhibits a sufficiently selective response for the target analyte. Under these circumstances, arrays and mathematical modeling, for example, principal components analysis, artificial neural networks, or other pattern recognition approaches, may be required [42],... 

---