Max amount of FITech students: 20
Persons without a valid study right at a Finnish university or university of applied sciences have preference to this course.
In this course, we aim to understand the basic mechanisms of Artificial Intelligence, so we focus on the fundamental principles. Artificial intelligence (AI) aims to understand thinking and intelligence in ways that enable the construction of computer systems that are able to reason in uncertain environments. Work in AI has supported the development of driverless cars and house-cleaning robots, as well as systems that have defeated world chess champions and planned space explorations.
The course has three core sections: search, representation, and uncertainty. In each section, it provides a thorough understanding of major approaches, representational techniques and core algorithms. In particular we focus on the trade-offs between the model structure of different frameworks and the algorithmic constraints that this structure implies. Central topics for searching include classical search algorithms, heuristics and relaxation, and adversarial game-playing. The representation section covers constraint-satisfaction, logical formalisms for representing knowledge, efficient algorithms for logical inference, and an introduction to planning. The section on uncertainty introduces probabilistic reasoning, the formalism of Markov decision processes, and an overview of Bayesian networks for modeling uncertainty. We conclude with an overview of generative AI and Large Language Models (LLMs).
Course contents
- What is Artificial Intelligence?
- Uninformed search, heuristics and adversarial search
- CSP, Local search, and optimization
- Markov decision processes
- Bayesian networks
- Markov models & hidden markov models
- Applications in NLP/vision/robotics
- Generative AI
Learning outcomes
Students completing this course will have an in-depth understanding of three core areas of AI and the connections among them, and with such other key AI areas as machine learning, robotics, natural language processing and multi-agent systems.
After the course, the students are able to
- choose the appropriate representation for an AI problem or domain model, and construct domain models in that representation
- choose the appropriate algorithm for reasoning within an AI problem domain
- implement and debug core AI algorithms in a clean and structured manner
- design and analyse the performance of an AI system or component
- describe AI algorithms and representations and explain their performance, in writing and orally
- critically read papers on AI systems.
Course material
- Stuart Russell and Peter Norvig, Artificial Intelligence: A Modern Approach, 3rd Edition.
- The course consists of lectures and solving exercises; no technology/programs other than Zoom and Moodle are needed.
Teaching schedule
At least one lecture and one exercise session per week, both possible via zoom. Teaching schedule can be found in Peppi closer to the course start.
Completion methods
Exam on campus and optional exercises.
More information in the Åbo Akademi study guide.
You can get a digital badge after completing this course.