Agent Architectures

An agent is a computer system that capable of flexible autonomous action. While Agent architecture is the model for software agents and intelligent control system. Basically there are 3 type of architecture which are:

1. Symbolic/logical (Proactive)
2. Reactive
3. Hybrid

Symbolic Architecuture

Symbolic Reasoning Agent

At the beginning, all the agents is designed were symbolic reasoning agents. It is the expressions that can propose to the agents to use the expression logical reasoning to decide what to do. 

To build an agent, there are exist a classical approach which is to view the system as a particular type of knowledge-based system.

This paradigm we called as symbolic AI, it contain the intelligent behavior can be generated in a system by giving that system a Symbolic representation of its :

• Environment
• Desired behavior
• Syntactically manipulating this representation

Symbolic is an architecture that :

• contains explicitly represented, symbolic model of world
• make decision via symbolic reasoning.

Keys Problems of Symbolic

1. The transduction problem which is to translate the real world into an accurate adequate symbolic description, in time for that description to be useful (Vision,speech,understanding,learning)
2. The representation/reasoning problems which is how to symbolically represent information about complex real-world entities and processes, and how to get angents to reason with this information in time for the result to be useful (Knowledge representation, automated reasoning, automatic planning)
3. Most of the search-based symbol manipulation algorithm of interest are highly intractable (Hard to control).

Deductive Reasoning Agent

Due to the problems of symbolic reasoning agent, a new method is use to build agent. The basic idea of deductive reasoning agent is to use logic to encode a theory stating the best action to perform in any given situation.

For example:

ρ be this theory (typically a set of rules)

Δ be logical database that describes the current state of the world.

Ac be the set of action the agent can perform

Δ ρϕ mean that ϕ can be proved from Δ using ρ.

/* try to find an action explicitly prescribed: best action to perform */

for each a ∈ Ac do
       if Δρ Do (a) then 
             return a
       end if
end-for
/*try to find an action not excluded: find an action that is consistent*/
/*with the rules and database*/

for each a  ∈ Ac do
         if Δρ ¬ Do(a) then
              return a
          end-if
end-for
return null /*no action found*/

The above algorithms explains the agent's behavior is determined by the agent's deduction rules (its programs) and its current database (representing the information the agent has about its environment.)

From the algorithm above, the for each a ∈ Ac do  , mean that the agent will keep loop the possible action inside the action list or array if the theory of the achieve the goals would achieve if do the following action, then it will return the action a, then the action will be done to achieve the goals,


Problems of Deductive Reasoning Agent

1. It was hard to convert the video camera to the real environment attribute, For example if the camera saw a dirt at floor, how the video camera convert the dirt to the value that reasonable to the agent.
2. Decision making assume a static environment: calculative rationality.
3. Decision making using first-order logic is undecidable.

Typical solution:

1. weaken the logic
2. use symbolic,non-logical representation
3. shift the emphasis of reasoning from run time to design time.