Intelligent Agent:
- Intelligent agents can be any entity object like human beings software on machines these agents can make decisions based on the input from the environment using its sensor and act on the environment using actuators ( screen, speakers.)
- AI enable agents collect input from the environment using sensors like cameras, Mics, or other sensing devices. these agents have abilities like real-time problem solving, error or success rate analysis, and Information retrieval.
- An intelligent agent can come in any of the three forms
- human agents
- robotic agents
- software agents
Human-Agent:
It uses the eyes, nose, tongue in other sensory organs as sensors to percept information from the environment and uses limbs and vocal tract as actuators to perform actions based on the information.
Software Agent:
It uses keypad stroke, audio commands as input sensor and display screen as actuators.
Robotics Agent:
Robotic agents use cameras and infrared radars to record information from the environment.it uses reflect motors is actuators l to drive output back to the environment . for example AI-based smart assistants like Siri Alexa use voice sensors to request the user's request and search for the reliable formation in the secondary sensors without human interference and actuators like voice and text module relay information to the environment.
These agents are classified into 5 types based on their capability range and extent of Intelligence-
- simple reflex agent
- model-based agent
- goal-based agent
- utility agent
- learning agent
SIMPLE REFLEX AGENT:
They are the basic form of agents and function only in the current state they have very low intelligent capability as they don't have the ability to store the past state. These types of agents respond to events based on predefined rules (which are pre-programmed). They performed well only when the environment is fully observable. Simple relex agents hold a static table from where they fetch all pre-defined rules for acting.
These agents are helpful only in a limited no of cases, like a smart thermostat.
MODEL-BASED AGENT:
It is an advanced version of a simple reflex agent. Like simple reflex agent it can also respond to events based on the pre-defined condition. It can store the internal state (past information ). It updates the internal step each step to perform any actions it relies on both internal state & current percept.
GOAL-BASED AGENT:
The action taken by these agents depends on the distance from the goal ( desire situation target). The actions are intended to reduce the distance between the current & desire state.
To attain its goal it makes use of a search & planning algorithm. Agent combine the provided goal information with the environment model to choose the action which achieves that goal.
Drawback:- They don't always select the most optimized point to reach the final goal.
The shortfall can be overcome by using a utility agent.
UTILITY AGENT:
The action taken by these agents depends on the end object. Utility agents are used when there are multiple evaluations of a problem and the best problem alternative has to be chosen based on each state`s utility.
LEARNING AGENT:
A general intelligent agent known as learning agent was proposed by Alan Turing & now is the preferred method for creating state of the system in A.I learning agent have learning ability so that they can learn from their past experience. These types of agents can start from scratch, can acquire significant knowledge from their environment. The learning agent has four major components which enable them to learn from past experience.
- Critics:- Evaluate how well the Agent performing ( performance benchmark)
- Learning elements:-n Take input from critics and help agents improve performance by learning from the environment.
- Performance element:- These components decided which action to be taken to improve performance.
- problem generator:- Takes input from other component and suggest action resulting in a better experience.
Rules
Few rules which agents have to follow to be termed, intelligent agents:
- Perception: The agent must have the capability to perceive information from the environment using the sensors.
- Observation: The inputs or observations collected from the environment should be used to make decisions.
- Tangible action: The decision made from the observation should result in some tangible action.
- Rational Action: The action should be rational.
Structure of Intelligent Agent :
The intelligent agent structure is the combination of agent function, architecture, and agent program.
AGENT= ARCHITECTURE + FUNCTION + AGENT PROGRAM
Architecture: It's the machinery on which an agent executes its actions. It's essentially a device embedded with actuators and sensors.
Agent Function: It helps in mapping all the information it has gathered from the environment into action.
Agent Program: The agent function performs the execution of the agent function.
Internal State Representation:
As the agent gets complex so does its internal structure.
Atomic Representation: In this case, the state is stored as a black box (i.e without any internal structure ) for example a robotic vacuum cleaner (Rambo) the internal state is a path already vacuum such representation works for a model and goal-based agent, and used various AI algorithms such as constrains satisfaction and Bayesian networks.
Factored Representation: The state in this representation is no longer a black box. It has attribute-value pairs also known as variables.
example: To find a route we have a GPS location. These add constrain to the problem, such representation works for goal-based agents and is used in various AI algorithms such as constrains satisfaction and Bayesian network.
Structured Representation: In this representation there is a relationship between variables and vector state - this induces logic in the algorithm. This high-level AI used in algorithms like first-order predicatace logic, knowledge base learning and natural language understanding.
All Topics:
Comments
Post a Comment