Fuzzy Logic: From Appliances to Intelligent Automation

Fuzzy logic has quietly arrived in our day-to-day lives. Take, for example, rice cookers.

Unlike regular rice cookers, which are either on or off, a fuzzy logic enabled rice cooker can consider the type of rice, room temperature, and other factors to determine the optimal temperature and time for cooking.

We see fuzzy logic in a broad range of applications: from washer-dryers and air conditioners to anti-lock brakes and traffic control. The concept has been around since 1965, when UC Berkeley professor Lotfi Zadeh came up with fuzzy sets theory. The theory deals with mathematical groups of elements. In normal cases, an element either belongs to a group or it doesn’t. Considering the set {cat, dog}. We can say the element “German shepherd” exists in the set, while a “rabbit” doesn’t. The membership of an element in a set is clear — it’s either ‘no’ or ‘yes’, 0 or 1.

Fuzzy logic, on the other hand, deals with what is in between — the ideas of “slightly,” “maybe,” or “probably.” A fuzzy set is one where elements do not have the straightforward characteristics of either being in the set or out of it. The membership function — whether or not something is in the set — is continuous between 0 and 1, where multiple factors are considered. This approach enables a machine to make judgment calls as a human would.

Fuzzy logic has also arrived in the office, helping to intelligently automate everyday business processes. At Automation Anywhere, we implemented fuzzy logic in IQ Bot, our offering for cognitive automation. IQ Bot uses fuzzy logic and other AI techniques to identify, extract, and categorize unstructured data from office documents — invoices, purchase orders, contracts, and others – so that data can be further processed.

In one of our earlier articles, we discussed how Optical Character Recognition, or OCR, is the foundation for cognitive automation. Popular OCR engines provide high accuracy and high speed, but still suffer from challenges like how to distinguish an “o” from “0” or a “5” from “S.”

Fuzzy logic removes the limitations from traditional OCR. In processing patterns and text in office documents, fuzzy logic can be used to describe the variability. The deformity of characters can be interpreted as a fuzzy membership function representing the degree to which a character should be considered in a set. IQ Bot, for example, conducts phonetic algorithm and fuzzy string matching against enterprise applications to validate and enrich extracted data. By combining detection and fuzzy logic recognition, it allows retrieval of documents with a high degree of accuracy.

Our world isn’t black-and-white, but a range of gray — from trying to cook the perfect meal to recognizing what data to extract from documents. Fuzzy logic enables us to automate the processing of these grey areas, freeing more of time to focus it on the things we value at home and work.

You can try out IQ Bot and see how it uses artificial intelligence techniques, such as fuzzy logic, by downloading Automation Anywhere’s Community Edition.