1

Artificial Neural Networks

According to Efthymiou & Aretoulis (2021), the history of Artificial Neural Networks

began in 1943 with Warren McCullloch and Walter Pits in 1943. They began by developing a

computational model for neural networks based on algorithms called threshold logic. Through

the model, it provided an opportunity for research in two ways. The first approach focused on the

biological approaches while the second approach focused on the application of neural networks

to AI (Avetisyan et al., 2020). Through the 2 research models, it led to a work on nerve networks

and their link to finite automata. Finite automata are another word for finite-state machine. It is a

computation model of computation; it changes from one state to another in response to some

inputs, the change is referred to as transition.

Yasnitsky (2020) states that the behavior of state machines can be observed in various

devices of the current world that perform a sequence of predefined activities depending on

various events sequence. For example, vending machines which dispense products when a

current combination of coins is deposited. Another example is elevators, their sequence stops by

the determination of floors requested by riders. Lastly, another example is that of traffic lights,

they change sequence when cars are waiting. Artificial neural networks can be defined as

computing systems that are generally inspired by the biological neural networks that constitutes

an animal brain. A neural network is a population of neurons that are interconnected by synapses

to carry out a particular function when activated. The key aim of the following presentation is to

discuss Artificial Neural Networks, Robotics and machine learning.

Understanding Artificial Neural Networks, it is essential to understand the complex

relationship between the brain and neurons (Barrett et al., 2019). According to Hasson et al.

(2020), Artificial Neural Networks is a collection of connected artificial neurons as of the

After receiving the signal, the artificial neuron processes them and signals the neurons connected

to it. The output of each neuron is computed by some non-linear functions of the sum of its

inputs. The connections between neurons are called edges which adjusts as learning proceeds

(Yang & Wang, 2020). Different layers of neurons perform different transformations based on

the inputs they receive. Learning strengthens neurons connection at the edges, transmissions

travel from the input layer to the output layer.

According to Qi et al. (2019), Artificial Neural Networks are trained through processing

examples which contain input and results data structures. Artificial neural networks learn to

perform tasks by considering examples without being programmed with task specific rules. The

network is presented with input-output pairs, and it adjusts its parameters to minimize the

difference between the predicted output and the actual output. Over time, the network learns to

generalize to new examples. This allows it to perform tasks, such as image classification, without

being explicitly programmed with task-specific rules. For example, when an image such as a cat

is used, an image recognition can learn by obtaining images that contain a cat, using the

examples, the neural network can identify a cat in other images. They do this without having any

knowledge about cats, they utilize identifying traits from examples that they process.

Artificial Neural Networks have various applications in different disciplines. It is due to

their ability to reproduce and model non-linear processes (Qiu et al., 2020). Artificial Neural

Networks are used in system identification and controlling processes, pattern recognition, face

identification, 3D reconstruction, data mining, visual recognition, text recognition, machine

translation, visualization, e-mail spam filtering and even clinical diagnosis. For instance,

Artificial Neural Networks have been utilized to diagnose various cancer types. They are also
able to differentiate various cancer invasive cells utilizing one cell information. Neural networks

have also been used in analysis of infrastructures for disaster management and predict the

settlement of a foundation.

Artificial neural networks have also been widely used in cyber security (Ali et al., 2021).

They are used objectively to discriminate between malicious and legitimate activities. For

example, classifying android malware and detecting domains that pose a threat. Currently, there

is research undergoing on how Artificial Neural Networks can be used in penetration testing. The

key objective is to detect credit card frauds, network intrusions and detect botnets. Artificial

Neural Networks are also used to make predictions and decisions. They can also be tailored at a

specific task at hand.

Ali, S. M., Elameer, A. S., & Jaber, M. M. (2021). IoT network security using autoencoder deep

neural network and channel access algorithm. Journal of Intelligent Systems, 31(1), 95-

103.

Avetisyan, B. R., Druzhinina, N. S., & Daudov, I. M. (2020, July). Neural networks and artificial

intelligence as trends for the development of the future. In Journal of Physics:

Conference Series (Vol. 1582, No. 1, p. 012005). IOP Publishing.

Barrett, D. G., Morcos, A. S., & Macke, J. H. (2019). Analyzing biological and artificial neural

networks: challenges with opportunities for synergy?. Current opinion in

neurobiology, 55, 55-64.

Efthymiou, K., & Aretoulis, G. (2021). Artificial Neural Networks for Predicting Conventional

Cost of Industrial Construction Projects.

Hasson, U., Nastase, S. A., & Goldstein, A. (2020). Direct fit to nature: An evolutionary

perspective on biological and artificial neural networks. Neuron, 105(3), 416-434.

Qi, X., Chen, G., Li, Y., Cheng, X., & Li, C. (2019). Applying neural-network-based machine

learning to additive manufacturing: current applications, challenges, and future

perspectives. Engineering, 5(4), 721-729.

Qiu, J., Wang, B., & Zhou, C. (2020). Forecasting stock prices with long-short term memory

neural network based on attention mechanism. PloS one, 15(1), e0227222.

Yang, G. R., & Wang, X. J. (2020). Artificial neural networks for neuroscientists: a

primer. Neuron, 107(6), 1048-1070.

the future. Current hypertension reviews, 16(3), 210-215.