Summary multi-agent systems course of pre-master data science and society. Elaborate summary including theory and practical information about the course. The practical information consists of how to apply the theory in NetLogo.
What is an agent: an entity that has perception-action capabilities? It can sense its
environment and act in it:
- Humans
- Cells
- Software
- Robots
Example of an agent:
Many types of agents that range from not intelligent to very intelligent.
Multi agent system: is scaling up from one agent to many agents. Modelling multi-agent
systems to try to understand how they work.
What are multi-agent systems?
- A system or group of (potentially) interacting agents
o In some environment that they can sense and act in
o Can communicate and solve problems together.
- Can form the bases for distributed AI systems
- We also model multi-agent systems that exist in nature to try to understand how
they work (ants, birds, economies, etc.)
- Whole (system) is greater than the sum of the parts (agents).
Examples of multi-agent system:
- Natural selection
- Smart home: has different sensors that can change the environment and they don’t
have all the information.
- Classroom: there are goals and sub goals.
- Blockchain: decentralized.
- Cellular/organs
- Insects colonies
- Humans
- Organizations/economies
- Video games
- Autonomous vehicles
- Drone swarms
,Something that is like a multi-agent system but does not meet the criteria of an agent:
- Surveillance camara: do not interact. If there was a human always watching, then it
would be a multi-agent system.
- Virus: if it is just a mechanistic case without sensing.
- A bunch of robots doing things together however, without sensing they are doing it
in a fixed time frame.
Characteristics of multi-agent systems:
- Agent design
- Environment
- Perception
- Control
- Knowledge
- Communication
Environment: agents have to deal with environments that can be either static or dynamic.
- Static: for example, if we are running an agent-based system is the environment
something fixed that does not change or are there also some parameter about our
environment that change? This ties into the agent’s capability to perceive not just
each other but also elements of that environment.
- Dynamics: MAS often dynamic (especially with learning): especially in multi-agent
systems we are more focused on an environment that changes.
Perception:
Inormation can be a lot of things, it can be something that occurs over time and you have
only acquired that information over a fixed time interval. It can also be something that
occurs spatially, at a particular point in the environment. This means that sometimes you
have partially observable systems where not all of the agents have all of the information
about what all the other agents are doing or what is going on in the environment. This
makes planning diffcult. So, in robotics, they do something like SLAM which is spacial
localized action mapping. It only has some perceptial capability as it is going throuhg the
environment and mapping it and then able to plan.
- Information is distributed in environment
o Spatially, temporally, semantically
- Partially observable
o Makes action planning challenging
,Control:
- Decentralized: a lot of agents have control over a few things about themselves but
there isn’t some governing control over the entire system.
o Robust
o Hard to divide decision-making
- Game theory: a general problem of multiagent decision making is the subject of
game theory.
- Coordination games: in a collaborative or team MAS where the agents share the
same interests, distributed decision making offers asynchronous computation and
speedups, but it also has the downside that appropriate coordination mechanisms
need to be additionally developed.
- Different control architectures and rules are possible
Knowledge:
- Levels of knowledge may differ
- Common or shared knowledge structures are important
o Knowing what other agents know
o Shared mental models, situation awareness, transactional memory system
Shared knowledge structures (sometimes called shared mental models): how much overlap
of this system of the situation do we have versus transactional knowledge which is more
about: I know that this person or this agent has this type of knowledge, so you know how to
retrieve that because you first know who has it to begin with. So, you are thinking about
what your agents needs to know (what kind of information is relevant to the agent).
Communication: how do the agents communicate? Sometimes this is just going to be
putting something out in the air (for example, ants release pheromones that other ants can
pick up). Sometimes it is going to be more involved than that and there is going to be more
updating. Particularly, when you need to reach a consensus or need to negotiate about
something. You need to have a protocol that is going to allow you to have agents to
somehow communicate.
- Two-way sender receivers
- Needed for condition and negotiation
- Protocols for heterogeneous agents: in order for exchanged information to arrive
safely and timely, and what language the agents must speak in order to understand
each other.
Applications:
- E-commerce, trading, auctions
- Robotics
- Computer games
- Social and cognitive science
- Internet
- Human-machine teaming
, Challenges:
- How can we understand and solve problems with multi-agent systems?
- How can agents maintain a shared understanding of their environments?
- How can we design agents that coordinate and resolve conflicts?
- What kind of learning mechanisms are there for agents?
- How can agents of different types interact effectively?
Lecture 2
What is agent-based modelling?
A model is an abstracted description of a process, object, or event. It does not completely
match up with the real world and the important thing is that it exaggerates certain aspects
at the expense of others. “Essentially, all models are wrong, but some are useful” (George
Box, 1987).
An agent-based model is a model that consist of agents.
An agent is an autonomous individual element with properties and actions in a computer
simulation. It can be a human, a company (in which case the agent-based model might be an
interaction between different companies).
Agent-based modelling (ABM): the idea that the world can be modeled using agents, an
environment, and a description of agent-agent and agent-environment.
Many agent-based models try to characterize behavior of an individual. It can also be done
on a general level not for one individual.
NetLogo:
Interface tab:
Every model in the model library has a setup and go button. And it is useful to have one in
your own model.
If you want to create a new button you go to button on the menu on the top of the screen
and choose a button. Click somewhere to place the button.
Besides the button we have slider (the density aspect). The slider has a global variable that
the slider is tight to. It has a minimum and a maximum and a increment (how big the steps
are) and what the default value is if someone loads it (current value).
Switch: it is on or off
Choosers allow you to set up variables that have a number of different choices.
The benefits of buying summaries with Stuvia:
Guaranteed quality through customer reviews
Stuvia customers have reviewed more than 700,000 summaries. This how you know that you are buying the best documents.
Quick and easy check-out
You can quickly pay through credit card or Stuvia-credit for the summaries. There is no membership needed.
Focus on what matters
Your fellow students write the study notes themselves, which is why the documents are always reliable and up-to-date. This ensures you quickly get to the core!
Frequently asked questions
What do I get when I buy this document?
You get a PDF, available immediately after your purchase. The purchased document is accessible anytime, anywhere and indefinitely through your profile.
Satisfaction guarantee: how does it work?
Our satisfaction guarantee ensures that you always find a study document that suits you well. You fill out a form, and our customer service team takes care of the rest.
Who am I buying these notes from?
Stuvia is a marketplace, so you are not buying this document from us, but from seller liekebuuron. Stuvia facilitates payment to the seller.
Will I be stuck with a subscription?
No, you only buy these notes for $5.08. You're not tied to anything after your purchase.
