FAQSearchEmail

humanlevelartificialintelligence.com   

  
 chapter3

Home | Videos | Contact Us   

 
Home
HLAI
UAI
Videos
Books
Patents
Notes
Donation

     
 

                     

                         << content                                      Chapter 3

Retrieving data from memory through intelligent pathways

Data in memory are stored using different data types.  These data types will come in the form of 5 sense objects, activated element objects, pattern objects or hidden objects.  Since the data in memory are different, how will the search functions work?  The answer is through intelligent pathways.  The robot has to learn how to analyze data in the real world.  Teachers have to teach the robot how to look at a form and extract information from the form.  For example, if the robot was asked a question such as:  “can you read me the name on the form?”, the robot has to look at the form, locate the name, and read what the name is.  Then, the robot has to face the teacher and tell her the name that was on the form. 

Using this lesson (an intelligent pathway) the robot can extract a form image from memory and use the lesson to extract the name on the form image.  This form image is data stored in memory, as an experience by the robot.   

In another example, a teacher plays a song for the robot and after the song is over, the teacher asks the robot to describe what category the song belongs to.  Next, the robot will analyze aspects of the song, such as rhythm and lyrics; and say to the teacher what the most likely category of the song is.

Using this lesson (an intelligent pathway) the robot can extract a song from memory and use the lesson to output the category the song most likely belongs to.  This song is a memory of the song, experienced by the robot.

The question is:  how does the robot know that the intelligent pathway learned by teachers can be used to analyze and extract information from data in memory?  The answer is through patterns.  FIG. 9 is a diagram depicting how the robot finds the patterns between lessons to analyze and extract information from data in the real world and analyze and extract information from data in memory. 

FIG. 9

In pathway23, the teacher teaches the robot how to analyze and extract information in a form he is currently looking at.  If the robot doesn’t understand the steps to analyze or to extract information, then the teacher has to teach the robot these steps. 

Intelligent pathway23 is universalized in memory, which means the robot can answer any question related to searching and extracting information on a form.  This can happen only if the robot was taught many similar examples.  For example, the universal pathway23 can answer similar questions such as: 

 

1.  “can you read me the phone number on the form” 

2.  “can you read me the occupation on the form” 

3.  “can you read me the home address on the form” 

 

The universal pathway23 can cater to similar questions.  The instructions to analyze and extract information are very similar. 

In pathway24, the teacher gives the robot a command:  “I want you to remember the form and read me the name”.  The robot’s brain will extract the memory of the form and use pathway23 to analyze and extract the name on the form.  Finally, he will tell the teacher what the name is, which is john doe

In pathway25, the teacher gives an ambiguous command to the robot; and based on logic (another intelligent pathway) the robot assume the teacher wants to know the name that was on the form.  Pathway23 has already been encountered and the robot answered the question in pathway23, he simply has to repeat that answer.   

To sum things up, the robot is taught many various examples and based on patterns and logic, he will know that pathway23 can be used not only for analyzing and extracting information on a form the robot is currently looking at, but also, to use pathway23 to analyze and extract information from a data in memory. 

 

Intelligent pathways to set a goal, plan a strategy and achieve the goal

The pathways in memory can create any form of intelligence.  Referring to FIG. 10, an intelligent pathway to set goals, plan a strategy and achieve goals can be created (called A1).  Teachers will have to teach the robot what are the steps to achieve a goal.  The steps are:  to set goals, plan a strategy, trial and error and successfully achieve goals.  During the planning and trial/error steps, there exists a loop because some strategies might not work and the pathway has to loop itself again to plan a new strategy and to take a different action. 

FIG. 10

 

FIG. 11

FIG. 11 is a diagram depicting an intelligent pathway called M1 that basically solves a problem.  The steps are presented in the diagram on how the robot will solve a problem.  Notice that intelligent pathway A1 and M1 are very similar.  In memory, these two intelligent pathways will be very close to each other because of their commonalities. 

Both A1 and M1 can be universalized so that they can cater to a wide variety of situations.  For example, the problem solving pathway M1 can be used for all problems.  It can be used to solve a business problem, a math problem, a science problem, a conflict problem, a personal problem, a videogame problem, or driving problem.  On the other hand, pathway A1 can be used to achieve “any” goal/s.  The robot can use A1 to solve a math problem or to drive a car to a destination or do a math problem, or do a series of math problems or past a videogame. 

Similar types of problems will self-organize.  For example, doing a science problem might be similar to doing a math problem or driving a car might be similar to driving a motorcycle.  Intelligent pathway A1 and M1 are universal pathways that will organize specific pathways in their hierarchical trees.    

Another note is that pathway A1 can encapsulate pathway M1 and/or vice versa.  Steps in pathway M1 might be to achieve 20 different separate goals.  Or steps in pathway A1 might be to solve 10 interconnected problems.  The brain of the robot will self-organize data in memory so that similar data will be grouped together.  A bootstrapping process occurs, whereby pathways in memory add, delete and modify data on previously learned pathways.

 

Intelligent pathways used to manage tasks, plan strategies and follow rules

We learned that intelligent pathways can search and extract data from memory.  These data in memory have different data types; and the storage of data are not preprogrammed by computer scientists.  The brain learns how to organize data through examples and lessons from school. 

Other remarkable things intelligent pathways can do for the robot are:  to manage tasks, plan strategies, extract relevant data and follow rules.  Referring to FIG. 12, the intelligent pathways the robot’s brain uses to take action controls the computer program in the conscious.  The optimal pathway controls vital things like what the robot’s goals are, what tasks to do in the future, what rules should be followed and what information in memory should be extracted for the purpose of logical deductions. 

The diagram in FIG. 12 really shows the overall idea behind the conscious and how the robot actually controls itself in an intelligent way to take action. 

 

The robot’s conscious applied to a videogame

The task of playing a videogame is a good example of how human robots act intelligently in a dynamic environment.  When playing a game, the robot has to use logic and common sense knowledge in order to past the game.  Playing a racing game or a RPG game is very difficult and the player has to understand the rules of the game, the objectives of the game, how the controls work according to the game, how to solve problems in the game, how to get the character from one destination to the next, how to come up with plans to beat the game and so forth. 

The legend of Zelda is a very good example because the robot has to use human-level intelligence in order to past the game.  The Zelda game isn’t like a side-scrolling action game, whereby the player accomplishes levels in linear order.  In Zelda, the player has to talk to characters in the game and these characters will tell the player what to do next.  If the player doesn’t follow the instructions from characters, he will get lost and will not past the game.  The key to passing the Zelda game is to use human logic to come up with planned strategies and to take action; and through countless trial and error.

The intelligence in the robot’s brain comes from a bootstrapping process, whereby new data is built on top of old data.  When we play a videogame, we are actually using our knowledge of playing a general game.  The lessons of playing sports games in real life, the lessons of playing chase master, the lessons of playing a board game, the lessons of driving a car, the lessons of an occupation actually comes from one universal way of playing a game. 

Referring to FIG. 13, pathway B1 is a universal pathway to play any game.  The steps in B1 are very general, in that, all games played have these linear steps.  If you observe a sports game or a board game, they have these general steps.  B2 is a more general pathway to play a game.  In this case, B2 represent playing a videogame.  All the intelligent pathways (B1-B3) are all encapsulated and structured in a hierarchical manner so that the data goes from general to specific.  Intelligent pathway B3, on the other hand, record detailed steps to play a specific game.  If the game is the legend of Zelda, the steps to playing this game are different from the steps to playing a racing game. 

  

In intelligent pathway B2, the player has to set goals in terms of the type of videogame played and based on the player’s goals in the game.  Next, in the videogame are various rules and boundaries the player has to follow.  Then, the player has to also know what buttons control what actions in the videogame.  When playing a videogame, the controller controls the actions of the character on the monitor the robot is seeing.  On the other hand, if the robot was playing a real sports game, he has to use his body to take action.

 

Knowledge of the videogame

When the robot plays a given game, the robot’s conscious extracts relevant data about the game (activated element objects).  The computer program in the conscious will open a valve and knowledge about the game will start to pour in.  The intelligent pathways will help a great deal in extracting the relevant knowledge, but the computer program in the conscious actually controls when certain rules might be needed or when certain knowledge is needed for logic in a given situation.  

For example, when playing a real baseball game, knowledge about what happens during the timeline of the baseball game will be mapped out by the conscious.  During the estimated timeline of the game, knowledge will enter the conscious at specific times.  In the beginning of the game, the player (the robot) will know where to go.  It will follow the commands given by the coach.  The coach will designate each player their positions and to set the batting lineup.  Next, the coach will tell everyone what the hidden signals are for this game.  The robot’s conscious will tell the robot where to be and what instructions from the coach he has to follow. 

When the game begins, the robot already has tasks it will do based on the lecture by the coach.  Previous knowledge about baseball will pour in such as where the player has to be and what its roles are.  The rules of the position the robot will play must be known.  These rules sets up the limited actions the robot can take, what rules to follow, what actions to take at given situations and so forth. 

As stated in the last chapter, the conscious primarily has 4 containers:  task container, rules container, planning container, and identity container.  Based on the position the robot is playing, the rules container is filled with rules (most notably if-then statements).  If the robot is playing center field, some of the rules might be:  If the fly ball is hit hard, back up; if the fly ball is hit lightly move forward; if catcher throws the ball to second base, move forward.  These are the rules that the computer program in the conscious is trying to maintain in the rules container.

As the robot changes positions, the rules in the rules container will be filled with their respective position rules.  The rules of a pitcher are different from the rules of a left fielder.  The computer program within the conscious will extract and maintain the rules in the rules container based on the tasks in the task container.                

Some games do not need to follow commands from someone.  Playing the legend of Zelda is a game that simply requires knowledge from memory.  The robot’s conscious will extract all the tasks and rules of a RPG game (short for role playing game) and at the beginning of the game, the robot knows what type of game it is, what its goals are and what rules to follow. 

This information has been learned through magazines and strategy guides of previously played RPG games.  The knowledge of what to do and how to play the game has been described in magazines.  The robot learned how to play an RPG game through reading videogame magazines. 

For me personally, when I play a new RPG game, my first goal is to understand the story and what the game is about.  Then, I would follow the guidelines from previous RPG games, which are:

 

1.  talk to all characters in the game and follow their instructions

2.  read the instruction manual

3.  keep a map of the game.  You will find a map either from the game or an instruction manual.

4.  past level 1 before going to level 2 and so forth.

5.  if you get stuck in the game, ask yourself why you can’t advance. 

 

These are just some important facts that are needed in order to play the legend of Zelda or any RPG game.  These facts will pour into the conscious the moment the robot wants to play a RPG game.  In fact, specific tasks/rules will pour into the rules container and task container at specific situations.  During the game, problem solving pathways such as M1 will be used to solve problems in the game or to come up with imaginative ideas to go around obstacles.  Other intelligent pathways will be needed during the game such as pathway A1 to achieve goals.     

 

<< content               next chapter >>

 

 

Home | HLAI | UAI | Books | Patents | Notes | Donation

Copyright 2006 (All rights reserved)