When utilizing traditional data mining techniques $($such as Association Rule Mining, Sequential Pattern Mining, clustering, and classification$)$ for various event detection applications, several significant limitations become evident. These limitations include the following:

Those techniques are designed mainly to focus on some data types, including IoT data.

Those techniques are typically assumed centralized and are computationally expensive.

Detection of various events $($such as damage, cracks, explosives, fire, etc.$)$is not tough with those techniques.

Those techniques are designed mainly to focus on transactional data.

Data patterns with those techniques are straightforward.