Artificial vs Rule-based Machine Learning.

AI refers to simple or complex activities in objects (machines) that mimics human intelligence. It is a concept of which machines that are programmed to take decisions like humans and perform actions such as learning or problem solving. Human intelligence is the innate ability of human to learn, understand or adapt, which vary from executing simple to complex tasks for reasoning, perception or decision making on various kinds of tasks. It is highly subjective on acquired knowledge of how to get things done or instructions for doing. AI mimics human intelligence to execute an action/task based on stored knowledge (or experience) gained from continuous doing of the same task or set of dynamic instructions to achieving some tasks. Let’s discuss further on how machines acquire knowledge to become self-autonomous.

Machine learning, deep learning vs rule-based AI

The goal of AI is to create technology that allows machines and computers to function like humans, in an intelligent manner [1]. Machine learning is a subset of AI which refers to a concept that machine automatically learns, adapt or improves its accuracy by experiencing new forms of data. Similarly, deep learning enables automatic learning through the usage of huge amounts of unstructured data such as text, images or videos [1]. Another form of AI is the rule-based (also known as production systems or expert systems) which uses rules as knowledge representation for defining systems that mimic the reasoning of human expert in solving a problem. The knowledge would be represented through if-then-else rules rather than procedural code. Rule based are the simplest and probably the oldest form of AI, embedded as expert systems which emerged in the 1970s and 1980s, and considered as successful forms of early AI [2]. The two major components of a rule-based AI are “a set of facts” and “a set of rules”, which tells what to do or what to conclude in different situations [3]. Machine and deep learning models typically require more data than rule-based models

Imagine if Robots were unbridled and can keep executing instructions based on continuous learnings from data ONLY. The infinite loop of learning for such a robot may become destructive beyond the scope of the originator. One common theme is the idea that machines will become so highly developed that humans will not be able to keep up and they will take off on their own, redesigning themselves at an exponential rate [1]

AI Strategy: Learning or rule-based models?

Machine learning and rule-based models have their own advantages and disadvantages. The approach totally depends on the situation and appropriateness for the development of business. The relevance of rule-based approach stems primarily from the specific knowledge about a narrow domain stored in the expert system’s knowledge base [5]. Unlike a machine learning approach, the system defines its own set of rules based on patterns it sees in data. The machine learning system constantly evolves and adapts based on training data streams and relying on models that use statistics. The internal working of a ML learning approach is that the internal workings of the system cannot be extracted, resulting in a black box and a lack of insight into how the system made its decision.

Do not make an error of presenting every AI task to machine learning models. Have you considered the nature of the problems to automate? The amount of data for the learning process and the type of output expected? Machine learning is better equipped to identify patterns in the data than asking people to both find the patterns and manually. Rules-based systems are best suited to applications that need lower volumes of data and very straightforward rules.

An author [4] suggests the best projects for rule-based models are when the output is needed quickly or machine-learning is seen as too error-prone. Rule based AI produces pre-defined outcomes and are usually domain specific and mimics the workings of an expert knowledge base in that domain area. The goal of a rule-based system is to capture the knowledge of a human expert in a specialized domain and embody it within a computer system using a deterministic approach as against statistical methods in machine learning. Its knowledge is represented by production rules. A production rule, or simply a rule, consists of an IF part (a condition or premise) and a THEN part (an action or conclusion). IF condition THEN action (or conclusion). Rule based AI system is limited by the size of its underlying knowledge base, thus implementing a “narrow AI” and less useful for solving problems in complex domains or across multiple simple domains

Applications of rule-based or Learning based approach

Grisenthwaite (Jeff Grisenthwaite, VP of Product, Catalytic) says machine learning is better equipped to identify patterns in the data than asking people to both find the patterns and manually develop rules for each of them [2].

Examples of learning approach AI are algorithms that predict real estate prices based on a review of historical sales prices and factors including location, square footage and amenities. Also finding patterns to detect and flag activity in banking and finance such as unusual debit card usage and large account deposits — all of which help a bank’s fraud department. AI deep learning approach is being used and tested in dosing drugs and administering different treatment in patients in the healthcare industry. A healthcare AI system called Deep Patient turned out to be very good at predicting disease, even such that were difficult for physicians to predict quite well. Grisenthwaite stated that machine learning beats rule-based approach in identifying patterns.

Examples of applications of rule-based AI include expense report approvals that define payment thresholds that require management approvals at various levels, or email routing that uses a list of keywords to determine the destination. Some systems combine rules-based with machine learning such as an advertising business that uses a rules-based system to search through a library of answers to prior questions on requests for proposal forms. The responses considered more relevant in that filtered library are then scanned by a machine learning algorithm to predict the best answer to each question. Combining rules-based systems with machine learning enables each approach to make up for the shortcomings of the other [2].

Key points

1. Machines that are programmed to take decisions, learn or solve problems just like human in Artificial Intelligence.
2. Deep learning enables automatic learning through the usage of huge amounts of unstructured data such as text, images or videos.
3. Rule-based AI captures the knowledge of a human domain expert within a computer system using a deterministic approach embedded in a set of production rules.
4. Machine and deep learning models typically require more data than rule-based models.

References

1. https://www.investopedia.com/terms/a/artificial-intelligence-ai.asp#:~:text=Artificial%20intelligence%20(AI)%20refers%20to,as%20learning%20and%20problem%2Dsolving.
2. https://www.aitrends.com/software-development-2/rule-based-ai-vs-machine-learning-for-development-which-is-best/
3. Grosan C., Abraham A. (2011) Rule-Based Expert Systems. In: Intelligent Systems. Intelligent Systems Reference Library, vol 17. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-642-21004-4_7
4. https://becominghuman.ai/the-key-differences-between-rule-based-ai-and-machine-learning-8792e545e6
5. http://www.umsl.edu/~joshik/msis480/chapt11.htm