Saturday, April 24, 2010

simulation of human intelligence

Topic: Simulation of Human Intelligence to an AI being
Contents:
• Abstract
• Introduction
• Simulation of human intelligence to an AI being – The idea of Sensory Toy
• Psychological reasoning
• Sensory actions
• Working of a Sensory Toy – A case study
• Conclusion
•Pictures
• graphs
• References

ABSTRACT:

Humans have contributed to a great extent to the meaning of life. The real conundrum is not when human life begins, but, 'what is the value of human life?' Humans are the key to existence. They have reflected their intelligence in discoveries and inventions which, by time, made them live a better, comfortable and happy life.
To create an artificial being has been a dream of man since the birth of science. By simulating human intelligence to an artificial being makes that being think and act like a human. How far we have come? The artificial being is a reality of perfect simulacrum, articulated in limb, articulated in speech, and not lacking in human response. Our aim is to make the artificial being to reach its highest form i.e., universally adopted mechanical organism, the basis for hundreds of models, serving the human race in all the multiplicity of daily life.
Why do humans need mechanical machines as slaves? And why does he want them to think like him? The answer lies in his perception. By simulation of human neural networks through algorithms, it is possible to replicate human intelligence, what we call “Artificial Perception”. But what about the human senses and human emotions? AI researches predict a possibility to simulate through advanced AI programs embedded into a sensory chip. LISP and PROLOG are some of the good AI languages which make this task possible.
The question of whether or not it is possible to simulate human emotions along with human intelligence has been perplexing scientists since the research of AI has begun long back in early 1950s. What is the result of this simulation? A near perfect ‘human-like' being which is designed to obey the three laws of Robotics. But there's a catch. Robots differ from AI beings in a way that AI beings are designed with human intelligence simulated through advanced DNA and Neuron sequencing technologies along with advanced sensor technology with which a robot is made of. In that case, we can say that AI beings are advanced forms of humanoids. However, humanoids look more human and AI beings behave like humans. They simulate the human responses and intelligence. However, simulating human emotions is a challenging task as they are not a result of any logic. Though these machines are made not to violate the three laws of Robotics, with the simulated human intelligence they may overcome barriers of these restrictions. it would be up to the designers of the Artificial Intelligence to specify its original motivations. Since the Artificial Intelligence may become unstoppably powerful because of its intellectual superiority and the technologies it could develop, it is crucial that it be provided with human-friendly motivations. A super intelligence is any intellect that enormously outperforms the best human brains in practically every field, including scientific creativity, universal wisdom, and social skills. This paper introduces a unique concept which surveys the design and properties of Sensory Toy which is responsible for the decision making of the AI being and its limitations. The aim is to design a sensory toy with intelligent behavioral circuits which makes logical decisions according to the situations overcome by the AI being. We, humans, can foresee a possibility of the result but cannot predict the consequences. God created Adam to love him back and so do humans!

INTRODUCTION:
This century is characterized by an orgy of research, discovery, and invention. Branches of knowledge, industries, social contexts, and technologies have appeared that could not have been anticipated. These developments are affecting everything from the paraphernalia of everyday life to ontological categories. Artists can establish a practice in which they participate at the core of this activity rather than as distant commentators or consumers of the gadgets, even while maintaining postmodern reservations about the meaning of the technological explosion .
A robot, to react to any given situation, should analyze its environment which we call “Task Environment”, which are essentially the “problems” to which rational agents are the “solutions”. In order to design an agent, initially we must specify the task environment as completely as possible. In addition to this, we must provide the following factors: Performance Measure, Environment, Actuators and Sensors. The range of task environments that might arise in AI is far-flung. That means for a given environment, the AI being can react in any way. The sensory toy of the agent provides innumerable ways to take an action or to react to a given situation. For example, if the agent is alone in the house, it may clean the house. This is a general human tendency to keep things better. Consequentially, the programs which are responsible for any new ideas for the being are given in such a way that they follow the human tendency and always check, the decisions they make, with the three laws of Robotics. The neuronal pathways present in the robot should be working with the human logic. But is it possible to categorize the environments? However, identifying the dimensions determine, to a large extent the appropriate agent designing and the applicability of each of the principal families of techniques for agent implementation.
1) If an agent's sensors give it access to the complete state of the environment at each point of time, then we say that the task environment is fully observable.
2) An agent need not worry about uncertainty in a fully observable deterministic environment. If the environment is partially observable, then it could appear to be stochastic. This is true if the environment is complex, making it hard to keep track of all the unobserved aspects.
3) In an episodic task environment, the agent's experience is divided into atomic episodes. Each episode consists of the agent perceiving and then performing a single action at a time. Many classified tasks are episodic.
4) If the environment can change while an agent is deliberating, then we say the environment is dynamic for that agent; otherwise, it is static.
5) The discrete/continuous distinction can be applied to the state of the environment, to the way time is handled, and to the percepts and actions of the agent.
The agent has to analyze the action that is performed after any given sequence of percepts. The ultimate goal of Artificial Intelligence is to design the agent program that implements the agent function mapping percepts to actions. Algorithms help in taking decisions for the AI being. However, algorithms are uninformed, i.e. they are given no information about the problem other than its definition. Intelligent agents are supposed to maximize their performance measure.
Despite their diversity, it is important to identify some common properties that agents exhibit, making the agents different from conventional software programs:-
• Agents act by definition autonomously on behalf of a user or a process, without the direct intervention of humans or others.
• Therefore, an agent contains some level of intelligence , ranging from pre-defined rules up to self-learning AI inference machines.
• This intelligence enables agents to act not only reactively, but sometimes also proactively .
• Agents have a social ability , i.e., they may communicate with the user, system resources and other agents as required in order to perform their task(s).
• Moreover, more advanced agents may cooperate with other agents to carry out tasks beyond the capability of a single agent.
• Finally, as mobile or even active objects, they may move from one system to another to access remote resources or even meet other agents and co-operate with them.

Simulation of Human Intelligence to an AI being –
The idea of Sensory Toy:

The idea, here, is to provide Artificial Intelligence to an agent so that it makes its own logical decisions by simulating human neural networks with genetic algorithms which is provided by a Sensory Toy which is embedded in the agent. The agent will take decisions given by the sensors provided with the sensory toy. Through learning patterns, by past experiences, the agent will be able to retrieve the stored actions or decisions and replicate them according to the situation. The learning curve is given in graph#1. The elements of the learning agent are: performance element and learning element. The performance element decides what actions to take whereas the learning element modifies the modified element so that it makes better decisions.
Design of learning element is affected by three major factors:
• Which components of the performance elements are to be learned.
• What feedback is available to learn these components.
• What representation is used for the components.
Each of the components can be learned from appropriate feedback. The field of machine learning distinguishes three cases: supervised , unsupervised and reinforcement learning. The protocols of learning are provided in the sensory toy of the machine. These will result in logical decision making analysis of the machine enabling it to take consequential actions.
The feedback is provided by the past experiences of the robot which will be stored in its buffer memory used to retrieve the actions when demanded by the situation. A logical agent typically has a goal and executes any plan of action that is guaranteed to achieve it. An action can be selected or rejected on the basis of whether it achieves the goal, regardless of what other actions might achieve; the machine's knowledge can at best provide only a degree of belief in the relevant sentences. The assumptions may be based on facts or learning experience.
According to the principle of maximum expected utility , the rational agent should choose an action that maximizes the agent's expected utility. This means that the agent has to calculate the various quantities, maximize utility over its actions, and away it goes. Though MEU principle defines the right action to take in any decision problem, the computations involved can be exorbitant, and it is sometimes difficult even to originate the problem completely.

Psychological reasoning:
A operational psychology has to be developed for artificial agents to use in reasoning about themselves and other agents. This approach is useful in framework of natural language understanding of the machine. This reasoning is construed by intelligent behavioral circuits which are present in the sensory toy which designs algorithms for the actions to be taken. However, this is certainly steadfast on the situation which the robot perceives. For example, if the robot is given a task to clean the house, its intelligent behavioral circuits will develop logical decisions to analyze the given problem and design an algorithm for the action to be taken. This problem will be interpreted by the sensory toy in such a way that the resulting decision will curtail the diversions. The most imperative cautions of the decision or action taken by the robot is that it should not clean the room that has already been cleaned by it. This is possible through psychological reasoning given by the sensory toy. We can say that sensory toy makes the AI being responsive of its environment and navigate the robot to come to logical conclusions.
Learning is the process of acquiring knowledge , skills , attitudes , or values , through study , experience , or teaching , that causes a change of behavior that is persistent, measurable, and specified or allows an individual to formulate a new mental construct or revise a prior mental construct (conceptual knowledge such as attitudes or values). It is a process that depends on experience and leads to long-term changes in behavior potential. Behavior potential describes the possible behavior of an individual (not actual behavior) in a given situation in order to achieve a goal. The reasoning done by the machine is through its perception. Simple introspection suggests that the failures of monotonicity are widespread in commonsense reasoning. It seems to us that, we humans, often “jump into conclusions”.
The process of learning in humans, however, is in a “natural” way. We lead our lives taking most of the things for granted. This “natural” intelligence is interpreted by the agent in its own way. This provides “awareness” to the given situation and act accordingly. The learning for the robot can be done in two ways:
1) When a task is given to the robot, the robot learns while taking decisions and performing actions,
2) When the robot does things on its own, i.e., it is not supervised by anyone or it is not given a particular task to complete, it learns on its own independent of the external decision making factors given by the sensory toy.
Sensory actions:
The decisions taken by the agent to perform an action is done by specifying sensory actions, which are categorized into two types; automatic sensing, which means that at every time step the agent gets all the available percepts and active sensing, which means that percepts are obtained only by executing specific sensory actions. Reasoning about the results of actions is central to the operation of a knowledge-based agent. The real world can be seen as consisting of primitive objects and composite objects built from them. By reasoning at level of large objects such as apples and cars, it is quite possible to overcome the complexity involved in dealing with vast numbers of primitive objects individually. There is a momentous portion of reality that seems to defy any obvious individuation – division into distinct objects.
Sensory actions are articulated in human responses examining the real world environment. The knowledge that supports its decisions is represented explicitly and can be modified. For example, in the real-time environment, it starts to rain, the AI being can update its knowledge of how effectively its brakes will operate; this will automatically cause all of the relevant behaviors to be altered to suit the new conditions. For the reflex agent on the other hand, we would have to rewrite many conditions – action rules. Goal-based machine's behavior can easily be changed to go straight will work only for a single destination; they must all be replaced to go somewhere new. Learning allows the agent to operate in initially unknown environments and to become more competent than its initial knowledge alone might allow. All AI beings can improve their performance through learning which is done in the sensory toy.
An AI being with several immediate options of unknown value can decide what to do by first examining different possible sequences of actions that lead to states of known value, and then choosing the best sequence. This process of looking for a sequence is called search. A search algorithm takes a problem as input and returns a solution in the form of an action sequence. Once a solution is found, the actions it recommends can be carried out. This is called execution phase. Therefore, the design of the agent would be a simple “formulate, search, execute” which would be performed in the sensory toy.
Representation of knowledge: Representation can be taken in two ways. It refers to conventions for structuring data in a computer program and also refers to the abstract structuring of expressions denoting facts. They are concrete and abstract representation. An abstract representation carves up this rarified domain into expressions, each denoting a different state of affairs. A concrete representation is a way of storing these expressions as data structures.
Gaussian convolution in the brain: Brain uses neurons to process information, and from what we call tell, one neuron passes a value to the next by changing the rate at which it fires. So the faster the neuron is firing, the larger the value it is trying to pass along. The visual cortex is that part of the brain where the information from the eyes ends up. A neuron there integrates information from several different spots on the retina, and thus different neurons will respond to different types of stimuli, occurring at different places on the retina.

Working of Sensory Toy – A case study:
Sensory toy is the core of the AI being. The decision making, action, simulation, etc are done in this system. Let us consider a scenario where the robot suspects an intrusion. The strength of a surveillance system depends upon the reliability of its algorithms. The quality of a system's algorithms can make a difference between a threat being detected as a critical alarm or a harmless spot of noise. The algorithm of intruder detection is given in picture#1. The graph of intruder detection through sound is shown in graph#1. The sensory toy detects any change in the environment and analyses the situation through its own predefined manner. It uses its own techniques for voice recognition and image processing.

CONCLUSION:
Human evolution has changed the phase of the world to a large extent. It might not be confounding if, in a few coming years of this embryonic technology, the robots would walk on the streets with humans. Human life would be more contented with artificial companions. The main rationale behind the creation of artificial life is that robots will never be hungry and do not consume resources beyond those of their first manufacture, were so indispensable an economic link in the chain mail of society. Human beings had created a million explanations of the meaning of life in art, in poetry, in mathematical formulas. Certainly, human beings must be the key to the meaning of existence. The equations have shown that once an individual space-time pathway had been used, it could not be reused.
AI bases its approach to creating real artificial intelligence on an unyielding philosophical basis. Rather than keeping our philosophy in the realm of theory, we apply it to our complete working process.
Sensory toy is an intelligent brain to the robot with neuron sequencing technology which would make it possible to re-erect the life artificially. This makes the robot to think and act judiciously and humanly. However, the robot will not be aware of its existence. This is because this so-called “awareness of self” happens in the subconscious mind which might not be attainable to simulate to the AI being. Paradoxically, this can be an upshot of the logical conclusions made by the sensory toy of the robot. Sensory toy is like an electronic scrounger that has arisen to hold the minds of artificial intelligence!!!

PICTURES:


picture#1
0 – call John, 1 – close the door, 2 – close the curtains, 3 - call police, 4 - inform the owner, 5 – ask him what he wants, 6 – switch on the lights, 7- ring the alarm bell, 8 – distract the intruder

GRAPHS:
Graph#1






REFERENCES:


• Stuart Russell, “Planning and acting in the real world” Artificial Intelligence – A Modern Approach, pp – 417 -440, Pearson Education Series in Artificial Intelligence.


• Charnaik, “Gaussian Convolution” Introduction to Artificial Intelligence, pp-110, Low price edition.


• Sensory actions: www.genextech.com

No comments:

Post a Comment