### Description

Answer 1:

Bayesian Network has a different probability approach that helps the users to prevent some of the issues interfering with the network nodes. According to the information, Bayesian Network nodes have variable domains that allow the IT users to understand the process necessary required ti solve the problem affecting network nodes (Feki-Sahnoun et al., 2018). Organization management should understand the effectiveness of the Bayesian Network and how it enhances the probability node management on the required conditions. The company management deploys the values of the nodes to estimate the domain classifier probability in the dependency leading to successful protection measures.

However, Naïve-Bayes is a network system with the effectiveness in classifying the nodes using a simple approach. The users focus on developing a Naïve-Bayes to contribute to a successful node identification while determining the conditional independence requirements (Li, & Abdul Rahman, 2018). Most information technology users deploy Naïve-Bayes to develop a conditional independence node while identifying their specification to limit the factors causing the complication on the node’s conditional independence. According to the comparison between the two networks, Bayesian Network is more complicated as compared to Naïve Bayes. Both networks are useful in network performances and it allows the users to understand the structure learning for the information analysis.

The IT experts should be assigned in developing a Bayesian Network model since it requires effective skills and knowledge to achieve the set goals. The IT users should create ecological management while considering useful diagrams for analytical reasons (Choi et al., 2017). The IT analysts should create better plans in creating an alpha-level BN model by involving the diagram for risk limitation. The company with an effective BN model should understand the structures that detect and prevent threats issues while focusing on conditional probabilities. Moreover, network system updates help the users to get protected from the increased security layers while restricting the intruders from causing network threats. The developed network models should be regularly updated to ensure there are improved network security layers during the BN development process to limit hackers’ threats.

**References**

Choi, Y., Darwiche, A., & Van den Broeck, G. (2017, August). Optimal feature selection for decision robustness in Bayesian networks. In *Proceedings of the 26th International Joint Conference on Artificial Intelligence (IJCAI)*. https://par.nsf.gov/biblio/10053970

Feki-Sahnoun, W., Njah, H., Hamza, A., Barraj, N., Mahfoudi, M., Rebai, A., & Hassen, M. B. (2018). Using general linear model, Bayesian Networks and Naive Bayes classifier for prediction of Karenia selliformis occurrences and blooms. *Ecological Informatics*, *43*, 12-23. https://www.sciencedirect.com/science/article/pii/S1574954117302042

Li, L. X., & Abdul Rahman, S. S. (2018). Students’ learning style detection using tree augmented naive Bayes. *Royal Society open science*, *5*(7), 172108. https://royalsocietypublishing.org/doi/abs/10.1098/rsos.172108

Answer 2:

**Naïve Bayes and Bayesian Networks**

Sharda et al. (2020) define Naïve Bayes as a simple probability-based classification method and a Bayes theorem derivative. It uses machine learning to solve classification type prediction problems. According to Sharda et al. (2020), the technique expects nominal values from output variables but assumes the input variables are independent. Bayesian networks are dependency structure representation tools that depict probabilistic relationships in multivariate models (Sharda et al., 2020). Sharda et al. (2020) add that Bayesian networks can model any system and reason with partially uncertain, complex, and non-linear contexts. Naïve Bayes and Bayesian Networks relate primarily in their foundation, Bayes theorem that deals with probabilities showing relationships between variables. More importantly, Bayesian network models will adapt to and model the complex structures established by naïve Bayes classifier. According to Sharda et al. (2020), the options for developing a Bayesian network are doing it manually, assisted by an expert, or analytically by understanding the network structure from available data. An expert may be unavailable or lack time to dedicate to model building. According to Sharda et al. (2020), a tree augmented naïve is an example of structure learning involving five steps. First, determine the conditional mutual information to understand the amount of information provided by a known class variable. Second, create a complete undirected graph, applying the conditional mutual function describing an edge’s weight. Third, create a maximum weighted spanning tree. In the fourth step, one chooses a root variable and converts the graph into a directed graph. Lastly, one adds a vertex and creates an arc connecting it to x_{i}-pairs (Sharda et al., 2020).

**References**

Sharda, R., Delen, D., & Turban, E. (2020). Analytics, data science. *& Artificial Intelligence: Systems for Decision Support. Eleventh ed: Pearson*.

Answer 3:

According to Sharda et al. (2020), the nine-step process in con-ducting a neural network project is **1.** Collect, organize, and format the data **2.** Separate data into training, validation, and testing set **3.** Decide on network architecture and structure **4.** Select a learning algorithm **5.** Set network parameters and initialize their values **6.** Initialize weights and start training/validation **7.** Stop training, freeze the network weights **8.** Test the trained network **9.** Deploy the network for use on unknown new cases

In step 1, the data to be used for planning and testing the organization accumulates. Huge examinations are that the issues friendly to neural association course of action and that sufficient data exists and can acquire. In step 2, planning data should recognize, and a course of action should make to test the organization’s show. In stages 3 and 4, the network designing and a learning strategy pick. Significant contemplations are the number of neutrons and the number of layers. In step 5, limits are for tuning the association to the ideal learning execution level. This movement introduces the organization’s heaps, and restrictions, followed by changing the guidelines as planning execution input is getting. Routinely, the underlying qualities are critical in choosing the sufficiency and length of the arrangement. Stage 6 changes the application data into the ANN’s sort and configuration. The application data depiction and requesting impact efficiency and, possibly, the exactness of the result. In steps 7and 8, planning is driven iteratively by presenting input and needed or known yield data to the ANN. The ANN registers the yields and modifies the heaps until the figured yields are excellent resistant to the data cases’ known works. At the point when the planning has total, it is crucial to test the association. In step 9, a precarious game plan of loads procures. Presently the association can reproduce the ideal yield given sources of information like those in the arrangement set. The association sets up to use an independent structure or an element of another item system where new information will acquaint it, and its yield will be a proposed decision.

**References:**

Sharda, R., Delen, D., & Turban, E. (2020). Analytics, Data Science, & Artificial Intelligence: Systems for Decision Support (11th ed.). Pearson Education.

Answer 4:

**The process Steps of developing Neural Network-Based Systems**

Neural network is a type of artificial intelligence that is inspired by the human biological concept. It does not however directly mimic the organic structure, but it seeks to conceptualize thought process when solving problems. There are 9 steps to building a neural network, however, this process may not be as granular as it sounds because there are other complexities involved.

**Step 1:** Data collection formatting and organization is the first step in setting up the network. Much like humans, the conceptualize a theory, one must have information of it. It is the same concept in neural network, the machine must be fed data and formatted into useful information.

**Step 2:** Data needs to be classified into subsets for the purpose of validation, training and testing.

**Step 3:** The network architecture needs to be tailored to expected outcome, as such, a network structure change is needed.

**Step 4:** Before results are expected, the machine must learn with provided data, this is where the algorithm crawls and gathers as much information as needed

**Step 5:** The parameters of the network must be configured, and values set to fit the network environment.

**Step 6:** Once step 1-5 has been implemented, the machine has to be refreshed for training and input validation to occur.

**Step 7:** Training needs to be stopped at this level, all network weights frozen

**Step 8:** Using chosen application, the trained network needs to be put to test, this is to confirm that results are as expected.

**Step 9:** In the final level, the neural network algorithm will be deployed into the field. However, this time, the cases deployed against may be unknown or totally new in its domain.

Reference:

Sharda, R., Delen, D., & Turban, E. (2021). *Analytics, data science, & artificial intelligence: Systems for decision support*. Harlow, England i pozosta?e: Pearson.