A Model of Artificial Networks - Essay Example

Synapses are essential to the transmission of nerve impulses and have, on many occasions, been associated with learning and memory. Synapses play a vital role in memory operation since they connect various neurons in the nervous network. The audio-visual centres of the brain work coherently to connect the voice we hear with the associated face. More and more synapses are formed that connect various nerve cells in the nervous network, thereby allowing a faster and more seamless flow of information throughout the neural pathway.

If we hear a similar sound in the future, our brain functions to display the images of the associated individual’s face stored in the memory (Jones, Frosbery, Taylor, and Gregory, 2007). This is how the biological system within the human body works to transfer information and make connections with different bits of related reports. Hopfield (1982) proposed a model of artificial networks comprising neurons, now referred to as the Hopfield networks. The mathematical or computational model is inspired by the biological workings of neural networks in the human body.

These artificial neurons are interconnected, forming a dense network consisting of N number of neurons, with weight wi and an output regulated till neural upgrading. A weighted sum is calculated using a value for the input xi. If the weighted sum is greater than 0 or 1, the output is ascribed a positive value, but if the weighted sum is less than 0, the result is explained as a negative value. The output status is maintained until it is upgraded again via synchronous or asynchronous updating.

The weight matrix plays a significant part in this as it represents the collection of weight magnitudes from nodes j to me of the neural network. In terms of logic programming, Hopfield networks are asynchronous. The neurons update their state deterministically (Sathasivam and Abdullah, 2008). In asynchronous updating, a neuron is selected, the weighted sum is calculated, and the output is updated. The selection of neurons for asynchronous updating can either be ordinal or random. Updating and synchronization are necessary for the operation of the computing elements. Each computing element may face latency due to the time it takes to activate the neurons. Networks can be built that take the delay into account; however, in the absence of such an assumption about delay time or latency in activation, the Hopfield networks use stochastic dynamics.

To sum up, the Hopfield Networks and the human brain present many similar characteristics, the most prominent one being the system of learning and memory. It can be agreed that learning is not merely the collection and retention of facts but rather an association of the various partial information we have in our minds. This helps humans connect with the audio-visual sensory impulses and give meaning to the information we collect from our surroundings. Similarly, the Hopfield Networks use logic programming to show unsupervised learning through associative memory.