Extending the AIDDL Common Library
Uwe Köckemann
2024-12-10
Overall Description
The AI Domain Definition Language (AIDDL) framework allows to create
integrative AI systems by assembling AI components in different ways. It
consists of various open source libraries that provide functionalities to the
framework. One of these libraries is AIDDL Common which is composed of
default implementations for many common AI problems and algorithms which
can be used for rapid prototyping of integrative AI systems.
The aim of this project will be to extend the range of available methods in
the AIDDL common library in a specific area of AI (or related) and provide a
detailed study of relevant literature of the tackled area as well as a comparison
of implemented approaches over benchmark problems.
Details
Specifically:
• Study the state-of-the-art of the selected area and write a literature survey
• Precisely define the problem that is solved
• Select and implement at least three methods
– Provide abstractions where useful (i.e., traits of interfaces that can
be re-used or capture common functionality)
• Write necessary type definitions in the AIDDL
• Create a series of benchmarks problems
• Evaluate and compare the methods on the created benchmarks
• Interpret results with references to existing literature where possible
• Document implemented methods
Suggestions
The following is a list of possible areas that students can choose from (in
agreement with the supervisor):
• Scheduling
• Probabilistic reasoning
• Probabilistic planning
• Logical reasoning
• Machine learning
• Constraint processing
• Combinatorial optimization
• Optimization
• Reinforcement learnin
Some of these areas are very wide and will need to be narrowed down for the
purpose of the project. In some cases, basic algorithms are already implemented
in the AIDDL framework so these should not be chosen for the project.
Good work may be merged into the public AIDDL repository on github
(aiddl.org).
Requirements
Technology:
• Scala programming language
• AIDDL framework (see aiddl.org and docs.aiddl.org for work-in-progress
documentation)
Annonsuppgifter
Annonsör: Örebro universitet
Ansök senast: Löpande
Annonskategori: Examensarbete, praktik, uppsats
Intresseområde: Data och IT, Teknik och matematik
Kontaktperson: Uwe Köckemann (Forskare) uwe.kockemann@oru.se
Webbsida: https://www.oru.se/