<< content Chapter 3
Details of a predicted model
Each predicted model will have the variables: focused objects, peripheral objects, future predictions and aided software programs (FIG. 9). There is no clear standard what the future predictions will be. It really depends on what the predicted model is trying to predict. The team of virtual characters has to decide what output will be presented for their predicted model. These outputs are the data that is seen by parent and child predicted models or neighbor predicted models.
Focused objects are the objects that the virtual characters are primarily concerned with. Their job is to do predictions based on the focused objects. There are also peripheral objects that are considered minor objects, but the virtual characters might need these objects in order to make a better prediction. The lower level predicted models help to prioritize and limit the amount of objects that each predicted model has to work with.
The common knowledge container has lists of what predictions to output for a given predicted model. The virtual characters can use this knowledge to make their predictions. A software program can be created to better aid a user to view and manipulate the ranked predictions. For example, there might be functions to insert variables into the predictions to make it more accurate. Or there might be functions to modify certain aspects to get a better prediction.
The virtual characters might take individual software programs from multiple lower level predicted models and come up with their own software program that can manipulate different predictions.
The prediction tree is just an outline that structures important objects for a given prediction. This way, the virtual characters can use their time wisely by doing work on important objects only. The predicted models are also structured hierarchically so that teams of virtual characters can concentrate on limited amount of objects to analyze. Each team that does predictions has to work in a united manner. The goal is to predict different aspects of a prediction, simultaneously. These teams of virtual characters should act as one entity that is aware of all knowledge generated by the prediction tree.
FIG. 10 is a diagram depicting time dilation for the prediction tree. All predicted models have to be worked on simultaneously. However, the top levels have to wait for the lower levels to do their work first. Thus, the top level predicted models have slower time and the lower level predicted models have faster time.
Work should be distributed equally among the predicted models in the prediction tree. There is no point in predicting the final score of the football game when the 4th quarter hasnít been predicted yet. There is no point in predicting the future possibilities of the quarterback when the quarterbackís brain hasnít been predicted yet.
In FIG. 10, B1 (X node) is the predicted model the team of virtual characters are working on. They will look at data from its neighbors (black nodes) and using this data they will output their future predictions and aided software programs. B1 is only concerned with data from itself and its neighbors. Anything outside its neighbors should not be analyzed.
Predicted models in the tree will be added, deleted or modified as work is done. Predicted models that virtual characters think are not important will be deleted. Predicted models that are not in the tree will be created based on demand. Pre-existing predicted models can also be re-organized in a different part of the tree. Teams of virtual characters have procedures that they will follow to add, delete and modify predicted models in the prediction tree. As more work is done on the prediction tree, the predicted models are arranged in an optimal manner.
This optimal structure of the prediction tree will allow the virtual characters to concentrate on the most important objects to analyze and to output accurate predictions.
Hypothetically, letís say that B1 wanted to use data from E3. E3 is an unrelated predicted model and it is very far away from B1. The Team of virtual characters from B1 will create a new predicted model called S4 that has both aspects of B1 and E3. This S4 will be attached somewhere that has the closest predicted model. S4 will be attached to parent nodes as well as child nodes.
The key here is that if S4 isnít a very popular predicted model and very little people like to make predictions there, then that predicted model will be deleted. If teams of virtual characters agree that this predicted model is important, S4 will stay.
In another case, a pre-existing predicted model can be changed in terms of the teamsí goals, purposes and predictions. The team can state that the focused and peripheral objects are not accurate and therefore, they should be changed.
Teams of virtual characters will act like competing businesses
When the prediction tree is generated, each predicted model will be assigned to certain specialized teams. Each team of virtual characters has to register and define what their expert fields are. Some teams specialize in ocean currents and others specialize in analyzing atom interactions. Every single predicted model or prediction tree self-organize in memory and software can be created to assign teams of virtual characters to predicted models.
It is prudent to assign more than 1 team to a given predicted model because you want two or more teams to compete with each other in who can generate accurate future predictions in the fastest time possible. The common knowledge container has a list of teams and how they rank. This list will motivate each team to do better in the future.
Teams can also dictate what predicted models they prefer to work in. They can work on one predicted model and then jump to another predicted model.
FIG. 11 is a diagram depicting the behavior of the prediction tree as time passes. The predicted models in the tree will expand as more work is done. Notice that B1, E3 and S4 expand as more predicted models are added. The more work done on the prediction tree the larger the tree will become.
Working in an expanding prediction tree
For predicted model B1, as the prediction tree expands, there will be more neighbor predicted models. B1 can search for spaced out neighbor predicted models instead of close-by neighbor predicted models. FIG. 12 illustrates that if the prediction tree expands dramatically, B1 can search for limited spaced out neighbor predicted models.
It is desirable to search for limited spaced out neighbors because the information in its close-by neighbors are too similar. The team is concerned about the focused objects, but in order to have a better understanding of alternative possibilities, different information must be analyzed and not similar information.
The signalless technology and its role
The signalless technology basically collects information from sensing devices like cameras and microphones and uses the AI time machine to create a perfect 3-d map of the current environment. In terms of the football game, all electronic devices like cellphones, cameras, sonar devices and microphones are used to collect as much data from the environment as possible. This data is then processed by the AI time machine and the entire 3-d map of the football stadium is tracked atom-by-atom. No dangerous em radiation is ever used such as x-rays or gamma rays. The AI of the time machine simply collects as much data from electronic devices (even robot pathways) and it uses this information to map out the atomic structure of the current environment.
The signalless technology collects as much information from the environment as possible and it uses artificial intelligence to fill in all the missing pieces. In later chapters this subject matter will be described in detail. In this chapter a summary will be provided.
The AI time machine can encapsulate work done by teams of virtual characters. In the signalless technology, the job of the virtual characters is to take information collected by electronic devices like cameras and microphones and analyze the data for meaningful information.
A simple example is to track where someone is. If a person goes into a bank and the security cameras capture his image, that means the person is in the bank. A more complex form of tracking someone is to use logic to figure out where someone might be located. Letís say that a team of virtual characters are interested in tracking where 2 people are. Hypothetically, there are 2 people living in houseA and person1 loves to watch cartoons and person2 loves to watch game shows. One day a signal from the TV station was sent to the virtual characters stating that someone from houseA is watch cartoons. The virtual characters will assume that person1 is at houseA watching cartoons. On further investigation, a camera picked up person2 walking to his work place. The virtual characters will use human intelligence and assume that person1 is at houseA watching cartoons, while person2 is at work.
Once all the intelligent objects are tracked such as human beings, animals and insects, the next step is to track non-intelligent objects like buildings, bridges, houses, stores, malls and so forth.
Tracking intelligent objects is important because intelligent objects move and they donít stay in one area forever. Non-intelligent objects stay in one area unless they are moved by another object. It is important for the signalless technology to first track all intelligent and non-intelligent objects in the current environment.
Once this is done, the signalless technology will use artificial intelligence to find out all the hidden objects that canít be sensed by electronic devices, such as molecules, atoms, distant objects and so forth.
In order to find out where atoms are located, the signalless technology has to analyze em radiation (from all spectrums) and to assume the existence or non-existence of atoms in the current environment. Also, movements of wind and sunlight can be used as data to find out hidden objects. For example, the pathways of em radiation can tell the virtual characters what objects the em radiation bounced off in order to reach the camera. These bounces create a map of the environment. Wind movement is also one way to find out how air travels and bounces off hidden objects.
Or the virtual characters can use spectrum analysis and human intelligence to guess what type of atom transmitted the em radiation and where this atom is located in 3-d space. For example, if you go to a place near a nuclear power plant, the camera will pick up radioactive matter in the air. This radioactive matter came from a power plant close-by.
In another example, the virtual characters can analyze a video and guess what place it is in the world. For example, if there is a camera that shows a house, the virtual characters can look at objects in the house to assume where this house is. The virtual characters can point to the hand bag and say, that hand bag is only sold in Korea. This indicates that the camera is probably located somewhere in Korea.
The team of virtual characters has to use a combination of methods described above in order to map out the current environment atom-by-atom. In the case of the football stadium, the signalless technology has to collect information from electronic devices, like iphones, ipads, computers, laptops, cameras, microphones and so forth, and use the AI time machine to process all that information. The desired output from the AI time machine is a perfect atom-by-atom map of the football stadium.
While the signalless technology is processing a map of the current environment, information will be sent to the prediction internet as soon as possible. The virtual characters working in the prediction internet will take that information and use it to make predictions.
There should exist an automated feeding system that gives data from the map to the appropriate predicted models. For example, if one predicted model is to predict the physical body of the quarterback, then the data regarding the quarterbackís physical structure is sent to that predicted model. In the lower levels, there might be a predicted model that predicts only the QBís left arm. The data from the map regarding the QBís left arm will be sent to that predicted model.
In another case, the entire map of the current environment is sent to the prediction internet and any virtual character that needs information from the map can have access to the information.
Predicted model outputs
All virtual characters have to understand that information from any of the predicted models changes constantly. Each team should be given notices of when the next modification will be available. Outputs from predicted models should not be based on only the most specific prediction. The output should be structured hierarchically Ė meaning the information is organized from general to specific. Other teams should be able to extract a general prediction or a specific prediction. For example, if one predicted model is to output the throw of the football, the future predictions can have 3 possibilities at the top (a general prediction) and 10,000 different future predictions at the lower bottom (a specific prediction). These predictions are ranked and probability statistics are included.
The difference between a general prediction and a specific prediction is that the general prediction has a higher probability of happening.
Merging of two or more predicted models
Letís say that a team of virtual characters wanted to merge multiple predicted models together and create a hybrid predicted model. They can use the AI time machine or a fixed software program that will generate the hybrid predicted model. FIG. 13 is a diagram depicting an example of a hybrid predicted model. The team of virtual characters are trying to merge three separate predicted model: 1. the quarterback and the receiver. 2. the coaches and referees 3. fans in the stadium. Each predicted model has been worked on and future predictions are presented.
The team of virtual characters will use the AI time machine to generate a hybrid predicted model based on all three predicted models. Objects in each predicted model will be analyzed and the hybrid predicted model will have new focused objects and new peripheral objects. Hierarchical nodes will contain the strongest groupings between the three predicted models.
Iím assuming that there are no pre-existing predicted models similar to the hybrid predicted model.
The next thing is for the team to determine where this hybrid predicted model should be located in the prediction tree. The hybrid predicted model has to attach itself to parent predicted models as well as child predicted models. It should also be located in an area where there are similar predicted models. These things can be accomplished by the AI time machine or by fixed software programs.
Merging multiple prediction trees
Letís say that the entire prediction tree of the football game has been predicted and future events of the football game are known. The team of virtual characters might want to combine multiple prediction trees. FIG. 14 is a diagram depicting the merging of 3 prediction trees. These prediction trees are: 1. the football game. 2. the hot dog stand outside the stadium. 3. the blimp above the stadium.
The hybrid prediction tree must establish important object groupings between the three prediction trees. All objects are prioritized as well. For example, the football game is very important because that is where most of the intelligent human beings are located. The hot dog stand is non-important and really doesnít affect the football game nor the blimp above the stadium. The blimp does in some minor way affect the human beings in the football stadium because they see the blimp in the sky and sometimes human beings focus their attention on the blimp.
The team of virtual characters will probably use the AI time machine or fixed software programs to generate the hybrid prediction tree. In the prediction internet, there are many prediction trees, and software programs can be designed to compare separate child prediction trees and extract a parent prediction tree. The parent prediction tree should be the optimal tree that contains hierarchically structured object groups between the three child prediction trees.
The priority of each prediction tree is very important. In the diagram, the football game has a 75% priority rate, the hotdog stand has a 5% priority rate, and the blimp has a 20% priority rate. This means that more prediction time should be devoted to the football game than any of the other two prediction trees. Itís about isolating objects, events and actions. The hotdog stand doesnít affect the football game (only at a microscopic level). However, the hotdog stand is affected by the football game. When fans cheer, the hotdog stand can hear the sound. If the hotdog stand sells 10 hotdogs instead of 9, the football game wonít be affected.
In some ways, all three prediction trees have relational links to each other and each can affect the future outcome. During a gameplay, the quarterback might be distracted by the blimp in the sky and he misses a throw to the receiver. This example shows that the blimp caused the quarterback to miss a throw to the receiver.
The relationships between objects will most likely be 5 sense data from human beings. The relationship between human beings in the stadium and the blimp will be the visual image of the blimp in each human beingsí eyes. There are very few relationships between the 5 senses of the human beings in the stadium and the hot dog stand because the people in the stadium canít sense anything from the hot dog stand. A fan in the stand might think about the hot dog stand and wishes he can go there to buy a hotdog. This thought might change the way he will act. And this action might affect the players on the field.
Copyright 2007 (All rights reserved)