Certainty Factor in Artificial Intelligence

The latest inventions of computers that can do activities that normally require cognitive abilities, including recognizing images, speaking, making decisions, or the natural language, is referred to as Artificial Intelligence (AI). AI entails the creation of techniques and computer programs competent in learning from data, reasoning, and coming to conclusions based on its analysis.

Artificial intelligence (AI) is establishing itself as an emerging technology with numerous applications and purposes. AI is regarded as the next-generation automation technology on which people may rely. AI is employed in a wide range of industries, including audio recordings, video recordings, animation, entertainment, public transportation, medical care, and finance.

Artificial intelligence is divided into two categories: Strong and Weak or Narrow and General. A Weak or Narrow AI has been trained to do fundamental tasks, including pattern identification, image recognition, and language interpretation. Weak AI is confined to simple and basic activities that are often utilised yet require fewer resources in order to function and adopt knowledge.

On the other hand, strong or general artificial intelligence may be utilised in many tasks where the output has characteristics similar to that of a human mind. Deep learning, robotics, machine learning, neural networks, and other technologies are examples of this. These have qualities that might put human intelligence to the test. Weak AI has been embraced by many businesses, while strong AI is a long-term achievement to be employed; there still exists research underway that will take time to complete. When Strong AI is effectively implemented, it will be a tremendous advance in the technical area, perhaps leading to a variety of greater discoveries that no one could have foreseen.

There are several characteristics that are to be considered while interpreting the input in AI:

• Preciseness
• Completeness
• Understandable
• Easily Interpreted

These elements are responsible for obtaining the desired output from the machine. The basic purpose of introducing Artificial Intelligence is to teach machines to perceive and analyse things as people do. There are certain techniques to do this for Artificial Intelligence to be able to make judgements and reasoning. Fuzzy logic, Bayesian networks, decision trees, certainty factor, and artificial neural networks are a few examples. These are generally regarded as a subject of research to investigate how to incorporate human intelligence in computers since they play a significant part in the decision-making process in Artificial Intelligence.

Let us go through the Certainty Factor in Artificial Intelligence in depth. It is essential for humans to be able to examine and analyse the possibilities of outcomes in real life. There are occasions where the predicted results are unlikely to occur, and this defines Certainty, where we hypothesise about the outcome to occur. However, because computers cannot think and function like people, they require some way for understanding and estimating the likelihood of a choice occurring. So, to implement this into the systems Certainty Factor was introduced.

Certainty Factor in AI

The Certainty Factor is employed in systems that utilise Artificial Intelligence. The certainty Factor is a numerical value that represents the likelihood of an occurrence or statement being true. We can examine how probable the statement is to be true and estimate the possibilities of it being true or false based on this numerical number. An agent can use this value numerically to determine whether the event is true or false. The CF is used to quantify uncertainty. A various number of observations are examined to decide which is true or not.

Certainty factor values vary from -1 to 1, with -1 indicating entire doubt or assurance that the hypothesis/assumption is untrue, 0 indicating moderate belief, and 1 indicating complete belief or assurance that a thesis is true. Values ranging from -1 to 0 reflect different degrees of unbelief or uncertainty, whereas values ranging from 0 to 1 imply several different levels of belief or confidence.

The Certainty Factor is computed by adding the percentage of agreement along with the amount of disagreement to a certain proposition/hypothesis. The quantity of proof establishing a hypothesis is referred to as the measure of assistance, and the quantity of convincing proof denying that hypothesis is referred to as the amount of counter-support.

The agent applies the certainty factor for each input statement to determine if the statement or hypothesis is true or false. For each situation, a minimal certainty factor is calculated; this minimum value is used as the threshold amount. When we receive a minimal CF, we use it as the basic metric to determine if the assertion is true or untrue. For example, suppose the minimum value is around 0.7. In that case, it functions as a threshold value, with values less than that regarded as untrue and statements with certainty factor higher than what the threshold value considered true.

For example, in the medical area, the doctor might assign an acceptable threshold value to the patient's medical reports and health conditions. This figure will help healthcare providers determine whether or not to prescribe the course of therapy to them for an extended length of time.

Certainty factor is measured in two different ways:

• Measure of Belief
• Measure of Disbelief

These provide the agent with the probability of the hypothesis being true or false.

The Measure of Belief

As we know, the value of the Certainty Factor always ranges between -1 and 1, and the measure of belief is used to estimate the likelihood of the hypothesis taking place or being true. The Measure of belief is given by the formula- MB [H, E]

Where MB is the measure of belief, H is the provided hypothesis, and E is the evidence, the evidence simply refers to convincing statements that support the hypothesis's possibilities of being true or false.

When MB [H, E] =0, then it is considered as the provided hypothesis being false for the given evidence this means the hypothesis does not support the evidence. When MB [H, E] =1, then it is considered as the provided hypothesis is true and the provided evidence is supported by the hypothesis.

The Measure of belief always lies in the interval [0,1].

The Measure of Disbelief

The Measure of Disbelief is given by the formula- MD [H, E]

Where MD is the measure of disbelief, H is the provided hypothesis, and E is the evidence, the evidence simply refers to convincing statements that support the hypothesis's possibilities of being true or false.

When MD [H, E] =0, then it is considered as the provided hypothesis being true for the given evidence provided and the hypothesis supports the evidence.

When MD [H, E] =1, then it is considered as the provided hypothesis is false and the provided evidence is not supported by the hypothesis.

To calculate the certainty factor from the above measures MB [H, E] and MD [H, E]. This is possible for two conditions:

1. If MB [H, E] =1, meaning the hypothesis H is true for the evidence E then the measure of disbelief will be MD [H, E] =0. This means the certainty factor for the hypothesis and the evidence is CF [H, E]=1.
2. If MD [H, E] =1, meaning the hypothesis H is false for the evidence E then the measure of belief will be MB [H, E] =0. This means the certainty factor for the hypothesis and the evidence is CF [H, E] =0.

Certainty factor can be calculated by using the formula,

CF = (level of support - level of counter-support) / (1 - level of counter-support).

Let us take a quick look at the certainty factor with an example. Consider the following scenario: a patient goes to a hospital for treatment. After receiving the test results, it turns out the patient had two symptoms that might cause a disease. Suppose the symptoms are Symptom 1 and Symptom 2.

Now chances for having the disease are 36% if the patient is having symptom 1, and the chances for having the disease is 86% only when the patient is having symptom 2. Now to calculate the certainty factor we apply the formula. Here the level of support will be the chances of occurrence of the disease based on the symptoms. There is no counter-support in this case because there are only chances for the disease to happen. Using the formula, considering the patient having symptom B:

CF = (level of support - level of counter-support) / (1 - level of counter-support)

CF= (0.88 - 0) / (1 - 0) = 0.88

We know that the Certainty factor will always be between -1 and 1. Thus the percentage will be determined to fit the CF requirements.

This means the chances of getting affected by the disease when the patient is suffering from symptom A will be 36%. Certainty factor can also be used for multiple hypothesis and pieces of evidences provided.

This way, the certainty factor will be a helpful tool in artificial intelligence for handling uncertainty, collecting evidence, managing the evidence, and to interpret the hypothesis this provides transparency to the AI because this gives us a detailed elaboration on how the certainty factor is calculated.

Let us understand some of the applications of the Certainty factor in real-life implementations:

• Weather Forecasting: The Certainty factor may be helpful in the prediction of weather by taking some pieces of evidence, that are the records of the previous climatic conditions, and predicting the occurrence of either rain or storms in that region.
• Automatic/Electric Vehicles: The certainty factor is used in self-driving or autonomous vehicles. For example, if a car travels at 65kmph in autopilot mode, it must use the sensors to detect obstacles and other vehicles ahead of it. The following will be achievable by acquiring information from the sensors, deciding whether there are obstacles, and attempting to avoid getting into contact with it.
• Forecasting Disasters: Disasters such as earthquakes, floods, and tsunamis can also be estimated by the usage of the Certainty factor by collecting the data of the probability and estimating the time of impact of that specific disaster. By providing evidence of the chances of the events and then the calculation is done.
• Medical Diagnosis: A combination of the patient's symptoms and medical records, the Certainty Factor can assist medical professionals in making precise recommendations on the treatment and creating effective medication for the patient.
• Risk predictions in software products: Since the product has to be risk-free and error-free, the certainty factor might be useful in assessing the odds of dangers that the designer may experience during the product's development phase. Identifying the possibility of the error affecting the product's performance will be simple by examining the conditions.

There are many such uses and implementations of certainty factor by using artificial intelligence; these are only a few among those.

The Certainty Factor sure is a successful technique for managing uncertainties while making judgements based on the information. It offers the potential to enhance expert systems' functionality and effectiveness throughout a variety of sectors and activities.

Certainty Factor is an extremely useful AI tool for dealing with ambiguity, merging knowledge, managing contradictory evidence, and offering transparency. It is a versatile tool that can handle quantitative as well as qualitative information and may be applied to many different areas of applications. It is a strong tool for generating educated judgements based on available data, and it has the potential to improve the performance of expert systems significantly.