Personal Blog : My OUM(Open University Malaysia) journey starting from Jan 2011 till graduate.
Everyone can copy the contents in this blog and please leave some credit or a backlink to me... -Thank You-

Tuesday, March 11, 2014

INTELLIGENT SYSTEM DEVELOPMENT - CBIS3203

Assignment Question :
CBIS3203 - INTELLIGENT SYSTEM DEVELOPMENT

Agent-based technology is the popular choice for building certain industry-driven systems.
a.    In your opinion, what are the fundamental characteristics of an agent-based system? Explain your answer.
b.    Explain the differences between agent-based systems and conventional systems.
c.    Can an agent-based make mistakes? Justify your answers with relevant facts.
d.    BDI (Belief-Desire-Intention) model is an approach in developing agent-based systems. In your own words, discuss the meaning of BDI-model and the situations in which this model can be applied.

===============My Answer=============



1.0       Introduction.
Agent-based technology is widely used in many electronic devices that we use today. It covers the construction industry, medicine, transportation, education, research, military, and so on. Agent-based technology is also known as an intelligent agent. It is arguably the main tool as a driver / brain of the software. The main purpose is to replace human tasks that need to be taken continuously. In addition, agent-based also able to make decisions based on the solution options available. Quoted from website umich.edu, intelligent agent is defined as "An intelligent agent is a software that assists people and act on their behalf. Intelligent agents work by allowing people to delegate work that they could have done, to the agent software. Agents can perform repetitive tasks, remember things you forgot, intelligently summarize complex data, learn from you and even make recommendations to you." Another definition is "An agent is a computer software system to play Whose Characteristics are situatedness, autonomy, adaptivity, and sociability" (Charles R. Dyer, 2003). These two definitions are almost the same mean-based agent that helps people do their jobs 24 hours non-stop and can act and make their own decisions based on user input or the current environment.
The purpose of this paper is to describe an agent-based technology in more depth. Writing will include the characters of agent-based. The literature also includes the difference between the methods without intelligent agent as against the use of intelligent agent. Description is also made to identify possible errors that occur in agent-based technology or otherwise. The latter is based on one model of agent-based technology model BDI (Belief-Desire-the intention).
2.0       Fundamental characteristics
            Generally, the characters can be divided into two, namely internal characteristics and external characteristics. Internal characters include nature conducted by the agent-based technology. The external characters, which requires the involvement and interaction with humans, other agents, the environment or other resources to enable agent-based technology to accomplish the task set.


            2.1       Internal Characteristics
Internal characters to be described is learning, reactivity, autonomy and goal-orinted.
                         2.1.1    Learning
Agents have the ability to learn from past experiences and use them as appropriate. Examples of this characteristics of agent-based technology has been used in a storm or tsunami warning systems. Tremor recorded by the agent will be stored and analyzed to produce a range of ground shaking will be said to be harmless. And the agent will automatically warn the user about the danger.
                         2.1.2    Reactivity
Agents are able to react quickly based on existing environmental information. For example, in the medical field. Content of platelets and white blood red blood platelets measured using a specially designed machine. Humans cannot calculate the ratio of white blood with red blood faster than existing agents in the machine.
                         2.1.3    Autonomy
Agents can control all internal actions performed by him. It does not require support for the act. Agent takes input from the user covers the important decisions to be made. Generally, an agent acting on its own in the tasks set.
                         2.1.4    Goal-oriented
Agents have a very clear goal. Agents will affect the environment when necessary to achieve the desired goals. For example: In a computer game, if we play war games and computers are our enemies. We find that the computer always want to try to beat us in various ways. This shows the computer's main goal is to beat people. All this is carried out by agents who created the game.
                        
            2.2       External Characteristics
The exterior features also are communication, cooperation, mobility and character.
                         2.2.1    Communication
Agents always require interaction with the environment, such as humans, other agents and other sources of information to ensure that the task executed successfully.
                         2.2.2    Cooperation.
When faced with a complex task, the cooperation between agents can produce a solution more quickly and effectively. Ability and accuracy will be reduced if it is implemented by an agent.
                         2.2.3    Mobility
Each agent can browse through an electronic communications network available. So it will act alone or combined as necessary.
2.2.4    Character
Agents have the same character with human behavior. Agents will always learn and improve behavior obtained as capable.




3.0       Agent-based System vs. Conventional Systems.
As we know, the characters software agents showing properties in which they have an advantage over traditional software. The perception that we can do is, agent-based are always aware of the environmental changes that occur. Agent-based also perform the duties of the complex itself. In addition, agent-based which we also know as intelligent agents always learn through hindsight. They also communicate with users and other agents.  Conventional systems do not operate on its own and requires input from the user. Each task must be performed on the input from the user. It does not use a sensor that is sensitive to the environment. In summary, agent-based versus traditional programs listed in Table 1.0.
Agent-based System
Conventional Systems
Aware of the changing environment.
Never bothered to change the environment. This is because it relies on existing databases.
Perform tasks large and complex in itself.
Tasks large and complex to be monitored by the user.
Agent-based constantly gather information and learn through hindsight.
Cannot learn on their own. All knowledge is the result of input from the user.
Software agents communicate with each other among themselves and users.
Systems do not communicate with each other.
Auto Complete
Users Input
Table 1.0
For example, when we want to search through the website. Agent-based system more quickly to opportunities and results compared to conventional systems. I want to search keywords "mh370" through google.com website. I do not have to type the whole keyword. Agent-based technology will help when I'm typing just two letters only, "mh". Some keywords will be an option and I can continue to choose the keywords that I want.

Figure 1.0
Note in Figure 1.0, when I type the letter "m", the choice of keywords will be displayed. However, agent-based keyword does not give the option I wanted.

Figure 2.0
In Figure 2.0, the letter "mh" is typed and selection keywords as "MH370" was listed by software agents in the google.com search engine. With this, I can continue to choose the keywords that I want quickly without having to type the entire keyword.
                
Figure 3.0
And the result is as shown in figure 3.0 which displays a link on the keywords "MH370". Agents will also give more preference to keywords that people use the site from time to time. Most keywords will be at the top choices.
Compared to conventional methods of systems, I have to type the whole keyword "MH370" and keywords will be matched to a database search engine that is next display links that contain keywords word "MH370". Many lost time when I have to choose the right link. This is because the link will not be displayed in the list of links to the answers received. In other words, search engines simply look for keywords "MH370" in their database without giving the options would be wise for users to choose.



4.0       Agent-based system can make mistakes.
Taking into account the basic principles of computers on which electronic devices are not all perfect. And of course the agent-based system can also make mistakes. But mistakes are probably too small and risky and dangerous to humans.  For example, the failure of the water heater. Agents are not doing the job in order to heat water at a temperature of 90 degrees due to equipment malfunction. This is not a hazard to the user. However, it cannot output as expected by the user to obtain water at a temperature of 90 degrees Celsius.
While agent-based built by programmers who know all the code is written and tested, but the truth behind the code so that there are many bugs that cannot be detected. "You should assume that, no matter how carefully you have designed and built your simulation, it will contain bugs (code that does something different to what you wanted and expected)." (Gilbert, 2007). Gilbert's statement shows that each agent built by programmers are not perfect for all ages. As if it was perfect at the time it was created, but so the agent-based code that will have unforeseen drawbacks. This arises when the agents communicate with each other and the old agent cannot be being understood that the new agent code.


Figure 4.0
As shown in Figure 4.0. Agents interact with environments through sensors and effectors. Agents will learn or receive signals from the environment using a scanner. Input received will be processed and sent to the detector to produce a response back to the environment. Problems will arise if the agent-based fail to recognize the new environment. This makes the agent-based fail to analyze data and make decisions. Next is not sending a signal to the detector to respond. This is an example of failure that may occur on the agent-based.
Another example of the failure of the agent in the event of damage to the sensors or effectors. If the sensors received improper interpretation of the environment. Indirectly, the agent is unable to perform his duties better. Similarly, in the event of failure on the effectors.
5.0       Belief-Desire-Intention (BDI) Model
The BDI agent model is built on a simplified view of human intelligence. In it, agents have a view of the world (Beliefs), certain goals they wish to achieve (Desires), and they form Plans (Intentions) to act on these using their accumulated experience.  Agents that are written using the BDI model are at a level of abstraction closer to normal human experience. ( http://aosgrp.com/products/jack/documentation_and_instructi/agent_concepts_and_techniqu/what_is_bdi.html).  As the name suggests, the BDI agents take into account the aspects of belief, desire and the intention. BDI is a key approach in developing a multi-agent and agent.
            5.1       Belief
Belief is the fact that represent the agent believes the environment. For example, the camera system Automated Enforcement System (AES) will capture images of vehicles exceeding the speed limit based on sensors installed on the roads. However, the agent does not believe to receive the data captured by the sensor if no objects or movements that occur when the sensor sends a picture.
            5.2       Desire
Desire is a goal to achieve something desired. Each agent has a desire to be achieved. Agents also have some desire at one time will cause conflict. For example, AES agents have the desire to record two vehicles on the road that crosses the sensor simultaneously. Agents need analysis of both the vehicle and make a decision. Are both the vehicle violates the speed limit or is only one. This condition can cause the conflict to an agent when making decisions. At the same time, agents must receive signals from the sensor if there are other movements that cross the sensor.
            5.3       Intention
Intentions refer to existing commitments on the agents to achieve goals. Intentions cannot be in conflict with each other. Agent must be consistent in all the intention. For example, AES sensor was recording two simultaneous movement of vehicles exceeding the speed limit. Agent will make decisions that may be one vehicle or two vehicles being driven over the speed limit. But not both be driven slowly and do not exceed the speed limit.
6.0       Conclusion
Based on paragraphs written above, it can be concluded that the agent-based system is very important software in the present. It helped facilitate people perform everyday tasks more efficiently and quickly. The only maintenance and very minimum observations, agents can operate for 24 hours non-stop with almost no mistakes.
We can imagine if we have to close the road lights every day at 7 am and reopened at 7 pm by pressing the switch. A lot of manpower is needed for this task. However, with the availability of agents, all of this is done automatically every day at regular intervals using a sensor that is sent to the next agent is sent to the effector for action.




Refferences
Stuart Russell and Peter Norvig, c (1995). Artificial Intelligence: A Modern Approach. Retrieved from http://www.cs.berkeley.edu/~russell/aima1e/chapter02.pdf
Open University Malaysia (Eds.). (2011). Intelligent System Development.
Gilbert (2007). Errors and Artefacts in Agent-Based Modelling . Retrieved from http://jasss.soc.surrey.ac.uk/12/1/1.html
http://aosgrp.com/products/jack/documentation_and_instructi/agent_concepts_and_techniqu/what_is_bdi.html
http://cis.k.hosei.ac.jp/~rhuang/Miccl/AI-0/2012-AI-0-L2.pdf
 

Table Grade