Artificial Intelligence with Uncertainty

Deyi Li, Yi Du

• Publisher: CRC Press
• Year: 2017
• eBook ISBN: 9781315349831
• Topic: Computer Science
• Subtopic: Computer Science General
• Index: Computer Science

This book develops a framework that shows how uncertainty in Artificial Intelligence (AI) expands and generalizes traditional AI. It explores the uncertainties of knowledge and intelligence. The authors focus on the importance of natural language – the carrier of knowledge and intelligence, and introduce efficient physical methods for data mining amd control. In this new edition, we have more in-depth description of the models and methods, of which the mathematical properties are proved strictly which make these theories and methods more complete. The authors also highlight their latest research results.

Chapter 1
Artificial Intelligence Challenged by Uncertainty

During the rapid development of science and technology in the nineteenth-century Industrial Revolution, machines were employed to reduce manual labor or replace it altogether. In the twentieth century, advances in information technology, especially the advent of computers, produced machines that reduced or substituted for the work of the human brain, prompting the birth and rapid rise of artificial intelligence. In the twenty-first century, the widespread applications of the Internet and cloud computing, intelligent computing, sharing, interaction, and swarm intelligence have revolutionized people’s lives, indicating the arrival of the new era of artificial intelligence.

Intelligence can be defined as wisdom and ability; what is called artificial intelligence (AI) is a variety of human intelligent behaviors, such as perception, memory, emotion, judgment, reasoning, proof, recognition, understanding, communication, design, thinking, learning, forgetting, creating, and so on, which can be realized artificially by machine, system, or network.

Over the past 60 years, there has been great progress in AI. However, machine intelligence, built on the basis of certainty or accuracy, is severely limited by its formal axiom system, the precision of which is unable to simulate the uncertainty of human thought processes.

Thinking is the human brain’s reflection on and recognition of the essential attributes and the inner linkages of objective things. Language is the carrier and expression of human thought. Human intelligence differs from machine intelligence and even other forms of biological intelligence, in that only human beings are able to carry forward, through oral and written languages, knowledge accumulated over thousands of years. The uncertainty of intelligence is inevitably reflected in language, knowledge, thought processes, and results. Therefore, this book focuses on the study of uncertainty: its representation, reasoning, and simulation of human intelligence; on basic certainty in human intelligence; and how uncertainty challenges artificial intelligence.

1.1The Uncertainty of Human Intelligence

1.1.1The Charm of Uncertainty

In the nineteenth century, the deterministic science represented by Newton’s theory created a precise way to describe the world. The whole universe, as a deterministic dynamic system, moves according to definite, harmonious, and orderly rules that can determine future events. From Newton to Laplace, and then to Einstein, they depict a scientific world that is completely determined. Followers of determinism believe that uncertainty is only due to people’s ignorance of, not to the original look of things. If we are certain of the initial condition of things, we can completely control their development.

For a long time the deterministic view of science limited how people understood the universe. Despite living in a real world full of complex and chaotic phenomena, scientists only saw the predictable mechanical world with its structure and operational rules. They considered uncertainty insignificant, excluding it from the scope of modern scientific research.

According to the deterministic view, the present state and development of the universe were decided in its early chaotic stage. Things as large as changes in the world’s landscapes, or as small as personal ups and downs, were determined 10 billion years ago. Nowadays people disagree with this point of view because the reality of uncertainty is all around us. However rich one’s knowledge and experience, it is impossible to predict what will happen in life, for example, who will win an Olympic medal or how likely one is to win the lottery.

With developments in science, deterministic thinking has come up against insurmountable difficulties in an increasing number of research areas, such as the study of molecular thermal motion. With the multiple degrees of freedom that a large number of factors introduced, single-value deterministic theory is not only unable to solve the inherent complexity of the system, but the complexity of the system will also result in fundamental changes in things. Even though the full trajectory of particles and the interactions between them can be precisely determined, it is hard to know the exact behaviors forming the whole. Thus, in the late nineteenth century, Boltzmann, Gibbs, and others introduced the idea of randomness into physics and established statistical mechanics. Statistical mechanics states that, for a group of things, Newton’s laws can only describe a general rule; individual members of the group cannot be described definitively, but only in terms of their behavior, that is, their “probability.”

The emergence of quantum mechanics had a further impact on deterministic theory by showing that uncertainty is an essential attribute of nature. Quantum mechanics studies the movement of groups of atoms or particles. Because the particles are very small, observation, however it is conducted, interferes materially with the object it is observing. We cannot accurately determine the coordinates and momentum of the particles at a given moment. The more certain the coordinates of the particles, the more uncertain their movement, hence the “uncertainty principle” proposed by the German physicist Werner Heisenberg. The uncertainty of the objective world is not a transitional state caused by our incomplete knowledge, but is an objective reflection of the essential characteristics of nature.

Scientists have also admitted that although what we call scientific knowledge today is a collection of statements with different degrees of certainty, all our conclusions about science contain a degree of uncertainty. In the twentieth century, the philosopher Karl Popper described physical deterministic theory in his book Objective Knowledge: An Evolutionary Approach [1] thus: “I believe Peirce was right in holding that all clocks are clouds to some considerable degree—even the most precise of clocks. This, I think, is the most important inversion of the mistaken determinist view that all clouds are clocks.” Cloud symbolizes the chaos, disorder, and unpredictability of uncertainties, such as climate change, ecosystems, celestial movements, Internet computing, human mental activity, and so on. Whereas the clock is accurate, orderly, and highly predictable. The Cartesians believe that all our actions merely seek to convert the cloud into a clock. Those who believe in nondeterminism think that in the physical world not all events are determined accurately in advance, in every detail. Many scientists and philosophers agree with the view that uncertainty is the objective state of human cognitive ability.

Clouds formed by drops of moisture in the sky take on form if viewed from afar but are shapeless when viewed close to. Their form is elegant and changeable, showing different poses, sometimes like blossoming cottons, sometimes rushing down like water, light or dark, curled or stretched and carefree. They float in the air, coming together and parting, changing all the time, stimulating poetic imagination, hence the use of the term “cloud computing” to describe the uncertainties of the Internet.

The undulating contours of hills, winding coastlines, and rivers extending in all directions are not smooth but uncertain irregular shapes whose size, length, and area change according to the scale on which they are being measured. Once the scale is determined, the measured value can be determined; within a certain range, there is a power function relationship between the scale and the measured values. The length of a coastline can only be determined by the device used to measure it. This is the uncertainty of measurement.

A line group is a point when viewed from a distance, a three-dimensional sphere when looked at closer; it becomes a curve when you approach the surface, and a one-dimensional element when looked at more closely; a careful look shows a three-dimensional cylinder then a two-dimensional plane when looked at even more closely. If you examine a fiber of wool it is one-dimensional, which becomes a three-dimensional cylinder if you look closer. If you look at the Earth from outside the galaxy it is a point; seen from within the solar system the Earth’s orbit is elliptical; viewed on a plane the Earth is a two-dimensional surface. If you stand in the desert what you see will be completely different to what you see when standing on a small hill. This is the uncertainty of system dimension.

It is difficult to accurately represent the uncertainties of relationships between variables using analytic functions. Examples where x_i is a controllable variable with a specifiable value within a particular range, and y_i is a random variable with probability distribution include agricultural production and fertilizer (y₁, x₁); blood pressure and age (y₂, x₂); and strength and fiber length (y₃, x₃). There are many more variables or random variables that are not entirely independent of, dependent on, or associated with each other. This m...

Table of contents

• Cover
• Halftitle Page
• Title Page
• Copyright Page
• Contents
• Preface to the Second Edition
• Preface to the First Edition
• Authors
• Research Foundation Support
• 1 Artificial Intelligence Challenged by Uncertainty
• 2 Cloud Model: A Cognitive Model for Qualitative and Quantitative Transformation
• 3 Gaussian Cloud Transformation
• 4 Data Fields and Topological Potential
• 5 Reasoning and Control of Qualitative Knowledge
• 6 Cognitive Physics for Swarm Intelligence
• 7 Great Development of Artificial Intelligence with Uncertainty due to Cloud Computing
• Index