In an effort to get robots in computers to behave like computers collaboration is on between the diverse fields of artificial intelligence, expert systems and human machine interactions. Researchers are trying to apply the latest learning in the cognitive neuroscience to eventually build machines that don't wait for our commands but act independently having acquired the cognitive shells of humans.
In cognitive science, computers can be used in three ways: to simulate cognition for artificial intelligence, to enhance cognition by assisting human intellectual activity, and to help scientists understand cognition by testing theories on large amounts of data. These three approaches are not mutually exclusive, since specialists in any of these areas frequently adopt techniques designed for the others. This article surveys theories of categorization and reasoning in cognitive science that have been implemented and tested in computer systems. Most of the ideas originated long before modern computers were invented, but computers provide an opportunity for developing them in greater detail than was previously possible.
Cognitive systems also known as biologically inspired systems, require the design of "intelligent agents". An agent acts in a specified environment - a physical robot, web crawler, character in a video game or an automated system for medical diagnosis. An intelligent agent is one that takes appropriate actions towards achieving its objective. In other words, the agent must have the ability to perceive the environment in which it is operating and then decide what action to perform and carry out the action. To enable perception, cognitive system must be multimodal that is their traits should include visual, touch, speech, textual or linguistic capabilities. Based on the perception decision making takes place but this can vary too depending upon whether the agent has complete or partial knowledge of surroundings, whether it is acting independently or in tandem with or against other agents and so on.
The most important element of an intelligent agent is its ability to improve its performance over time, through learning. This is achieved by repeatedly performing 'sense-think act' cycle, in much the same way as humans do.
Cognitive systems or biologically inspired systems thus transform environmental inputs into decisions and subsequent actions (if any). The fundamentals of these systems are inspired by how the human brain performs this task. This is usually examined using data obtained from computer-controlled experimental tasks. As mentioned earlier, the general aim of a cognitive system is oriented towards developing hardware and computational infrastructure for development of novel neurocomputing intelligent architectures, based on the most recent evidence from neuroscience and cognitive science.
In an intelligent living environment, robotics will be the part of our lives. To have an intelligent robotic system, one needs to better understand the biologically-inspired processes by which motor action sequences are pre-planned in away that anticipates the future state of the robot's interaction with its environment.
Just like humans, the cognitive system I supposed to perform even If the habitat changes. If behavior fails, or a receptive field agent detects that sensor values are not consistent or reasonable, the sensor backbone is altered. The sensor backbone can then identify alternative schemes or behavior to replace the problematic behavior.
1. Computation in Cognitive Science
Theories of categorization in artificial intelligence, information retrieval, data mining, and other computational fields are no different in kind from theories that predate modern computers. The computer, however, introduces two important elements: it enables theories to be tested on large amounts of data, and it enforces precision, since no program running on a digital computer can ever be vague or ambiguous. Both elements can be helpful in formulating and testing theories, but neither can guarantee truth, relevance, or usefulness. Sometimes, as Lord Kelvin observed, precision can be a distraction: "Better a rough answer to the right question, than an exact answer to the wrong question." To avoid a bias toward answers that are easy to program, it is important to consider questions posed before computers were invented.
Let us survey a variety of computational methods that have been applied to categorization and related methods for reasoning with and about the categories. These methods can be used for three different purposes:
- Artificial intelligence: From the earliest days, computers were considered "giant brains", which had the potential to mimic and perhaps surpass human intelligence. Good (1965) predicted "It is more probable than not that, within the twentieth century, an ultraintelligent machine will be built and that it will be the last invention that man need make." Except for chess playing, that prediction has not come to pass, but attempts to achieve it have contributed to a better appreciation of the richness and complexity of human intelligence.
- Intelligence enhancement: Computer capabilities are very different from and complementary to human abilities. That difference has led to a wide range of tools that supplement human cognition: information storage, management, search, query, and retrieval; text editing, analysis, translation, formatting, and distribution; calculation in graphics, spreadsheets, statistical packages, and formula manipulation; and computer-aided human communication and collaboration.
- Hypothesis testing: Psychology, linguistics, and philosophy deal with complex phenomena that cannot be described by the elegant mathematics used in physics. Without computers, theorists are often limited to studying an unrepresentative sample of oversimplified examples or to collecting statistics that show trends, but not causal connections. A computer, however, can analyze large volumes of realistic data and test hypotheses about causal mechanisms that could generate or interpret such data.
These three approaches differ primarily in their goals: simulation, enhancement, or understanding of human cognition. Computational methods designed for any one of them can usually be adapted to the others.
The AI field has been able to sophisticated programs and devices with in depth expertise in specific fields. However, creating a program or robot that has basic human qualities like common sense, is still some way off. Such works in progress world wide are bringing us closer to that elusive goal of producing a truly cognitive machine that would mimic what's effortlessly achieved by an average 5-year-old human child!.
Related Online Articles:
No comment yet. Be the first to post a comment.