Intelligent Systems, Lecture 1:
What makes an intelligent system?
An intelligent agent:
- Cameras and microphones act as sensors
- Speech and human-like actions as actuators
Main intelligent methods:
- Perceiving
- Rational
What is trying to be achieved —> how can this be achieved
- Explicitly representing knowledge
- Good ad learning and adapting
Note: generic methods
Core topic 1: Problem solving:
Based on the notion of systematically looking at solutions, and also using internal knowledge of
humans.
Core topic 2: Knowledge and Reasoning:
On one hand we have an enormous amount of knowledge, that can also be used for network
structure (knowledge graphs), or even older classi cations schemes of animals species. We have
thousands of these sources that can be used for Ai applications. On the other hand we also have
logic.
Core topic 3: Adaptivity and learning:
Humans are good at identifying objects, we somehow have conceptual images, that we also want
computers to have. Intelligent systems should be able to classify di erent things. Machine
learning and linear regression. There are more ways to looking at an activity, how can we simulate
approaches in machine learning?
Schnapsen as a learning tool:
Schnapsen is a popular card game
It has an adversarial agent sitting (opponent interferes with your objectives. It is an imperfect
information games (there is an enormous uncertainty). Explicit knowledge available, we know the
rules of the game, and whether there is a strategy.
fi ff
What makes an intelligent system?
An intelligent agent:
- Cameras and microphones act as sensors
- Speech and human-like actions as actuators
Main intelligent methods:
- Perceiving
- Rational
What is trying to be achieved —> how can this be achieved
- Explicitly representing knowledge
- Good ad learning and adapting
Note: generic methods
Core topic 1: Problem solving:
Based on the notion of systematically looking at solutions, and also using internal knowledge of
humans.
Core topic 2: Knowledge and Reasoning:
On one hand we have an enormous amount of knowledge, that can also be used for network
structure (knowledge graphs), or even older classi cations schemes of animals species. We have
thousands of these sources that can be used for Ai applications. On the other hand we also have
logic.
Core topic 3: Adaptivity and learning:
Humans are good at identifying objects, we somehow have conceptual images, that we also want
computers to have. Intelligent systems should be able to classify di erent things. Machine
learning and linear regression. There are more ways to looking at an activity, how can we simulate
approaches in machine learning?
Schnapsen as a learning tool:
Schnapsen is a popular card game
It has an adversarial agent sitting (opponent interferes with your objectives. It is an imperfect
information games (there is an enormous uncertainty). Explicit knowledge available, we know the
rules of the game, and whether there is a strategy.
fi ff
