01whole.pdf (22.89 MB)
Statistical relational artificial intelligence for event recognition
thesisposted on 2022-03-28, 13:12 authored by Christopher Chan
Event recognition is one of the main areas of artificial intelligence research at present. It has many real world applications and syste4ms for event recognition are built around action logic and the forms of calculus that make up this logic. These systems are then implemented, commonly with logic or functional programming, with the exact syntax decide by the chosen calculus. Recent developments show a trend towards video recognition and improvement in this area. A successful implementation would be a combination of the event calculus and pobabilistic logic programming. We attempt to imlement a system that uses the event calculus and pobabilistic logic programming to learn the intangible rules that govern what effect events in the environment have on the environment and the agents residing in it. The objective of this project is to answer the question of how we can combine logic with probabilities, allowing us to learn the parameters and the structure of domain-dependent axioms in the event calculus.