Comprehensive coverage

The artificial intelligence that will reduce our risks

New research in the field of artificial intelligence and risk management may improve the accuracy of the risk assessment process and prevent errors resulting from the effects of the human factor

Image of risk matrix by artificial intelligence. Courtesy of Guy Burstein.
Image of risk matrix by artificial intelligence. Courtesy of Guy Burstein.

Risk management is an essential process in the field of project management, systems engineering, supply chain and production and many fields in the world of industry and economy as well as in the military field. The purpose of the process is to assess the probability that a certain failure will occur and, in the event of a failure, what is its effect on the project or system. Two studies conducted by Ph.D. student Guy Borstein and Dr. Yanon Zuckerman from the Industrial Engineering and Management Department at Ariel University offer a method for the application of neural networks for the purpose of locating patterns of risk factors and a way to reduce deviations and errors in the process.

The research paper published this week in the journal Risks deals with the methodology of breaking down the risk into secondary factors and recombining them in a fairly simple formula that is explained in the article, which enables a risk assessment of the probability of the event materializing based on several parameters such as: the complexity of the event, its frequency and variability, and other quantitative and qualitative characteristics that significantly reduce the deviation caused by cognitive and personality biases of the human risk assessor. The methodology was tested both by research on surveyors and in a simulation that simulates the process and showed a significant improvement in reducing the variability compared to the usual method in which risk is assessed. According to the researchers, one of the problems that the research tries to solve is the problem of lack of objectivity and getting different results in assessing the probabilities of failure by different risk reviewers testing the same system and the new formula may neutralize some of these biases.

In a previous research paper published in the Journal of Risk and Financial Management The researchers proposed the use of a neural network that is trained on the basis of data from existing risk surveys in order to give objective results that can replace or assist the risk surveyor in determining the probability of risk. The idea of ​​the researchers is to enable the neuron network to learn in supervised learning about each vector of risk factors what the expected risk is, according to a large-scale analysis of hundreds of surveys and audits conducted in many factories and companies.

According to the researchers, the methodology they developed is a tool that may be applied in many processes in which the wisdom of the masses of many surveys, tests and similar audits can be used in order to make the process more objective and scientific than it is today and at the same time to optimize the process and make it mechanized.

More of the topic in Hayadan:

Leave a Reply

Email will not be published. Required fields are marked *

This site uses Akismat to prevent spam messages. Click here to learn how your response data is processed.