Top

Published in:

2021 | OriginalPaper | Chapter

4. Artificial Intelligence

Author : Volker Lang

Published in: Digital Fluency

Publisher: Apress

Activate our intelligent search to find suitable subject content or patents.

search-config

AI-assisted search

Off

Abstract

Artificial intelligence (AI) is the most versatile digital technology introduced in this book and among the most frequently used buzzwords in media today. It is often used in conjunction with the related terms machine learning, neural networks, big data, and deep learning, which we will discuss in this chapter, too. We will see that artificial intelligence offers a wide range of applications across all sectors of modern business and society.

Dont have a licence yet? Then find out more about our products and how to get one now:

Springer Professional "Wirtschaft+Technik"

Online-Abonnement

Mit Springer Professional "Wirtschaft+Technik" erhalten Sie Zugriff auf:

über 102.000 Bücher
über 537 Zeitschriften

aus folgenden Fachgebieten:

Automobil + Motoren
Bauwesen + Immobilien
Business IT + Informatik
Elektrotechnik + Elektronik
Energie + Nachhaltigkeit
Finance + Banking
Management + Führung
Marketing + Vertrieb
Maschinenbau + Werkstoffe
Versicherung + Risiko

Jetzt Wissensvorsprung sichern!

inform now

Springer Professional "Technik"

Online-Abonnement

Mit Springer Professional "Technik" erhalten Sie Zugriff auf:

über 67.000 Bücher
über 390 Zeitschriften

aus folgenden Fachgebieten:

Automobil + Motoren
Bauwesen + Immobilien
Business IT + Informatik
Elektrotechnik + Elektronik
Energie + Nachhaltigkeit
Maschinenbau + Werkstoffe

Jetzt Wissensvorsprung sichern!

inform now

Springer Professional "Wirtschaft"

Online-Abonnement

Mit Springer Professional "Wirtschaft" erhalten Sie Zugriff auf:

über 67.000 Bücher
über 340 Zeitschriften

aus folgenden Fachgebieten:

Bauwesen + Immobilien
Business IT + Informatik
Finance + Banking
Management + Führung
Marketing + Vertrieb
Versicherung + Risiko

Jetzt Wissensvorsprung sichern!

inform now

previous chapter Blockchain Technology

next chapter Your Digital Action Plan

Technically speaking, chemical reactions generate a potential difference that is associated with an electric voltage. This voltage gives rise to weak electric currents that are transmitted by axons.

The Greek philosopher Plato already used the term cybernētēs in this sense to signify the self-governance of people in his famous literary dialogue “Alcibiades I” featuring a philosophical conversation between the two philosophers Alcibiades and Socrates.

The Logic Theorist was a computer program that mathematically proved 38 of the first 52 theorems in Chapter 2 of the famous Principia Mathematica, a three-volume work on the foundations of mathematics that has been published by Alfred Whitehead and Bertrand Russell in 1910.

Automata theory is a discipline in theoretical computer science that studies abstract machines and problems that can be solved by such automata specifically.

A nice online version of this worldwide first chatbot is available on www.masswerk.at/elizabot/, for example.

The “then” part of the programming code or statement basically tells the machine what to do for each of the “if’s” and “else’s.”

Checkers is a strategy board game for two players. It involves making diagonal moves of uniform game pieces (black or white) and mandatory captures by jumping over opponent pieces.

A nice documentary about the development of IBM Watson is available on YouTube via https://youtu.be/X0JZMHhupTs/.

Prediction is about analyzing information you have to generate insights and information you do not have. It is at the heart of making decisions under uncertainty.

There are in fact other optimization algorithms, too. They are called stochastic gradient descent, mini-batch gradient descent, momentum, and Adagrad, for example. Gradient descent is, however, used most frequently and is considered to be the standard method of choice.

A nice website to play with selected machine learning models can be found on https://teachablemachine.withgoogle.com/.

See www.image-net.org/ for further details.

See, for example, www.economist.com/technology-quarterly/2020/01/02/chinas-success-at-ai-has-relied-on-good-data/.

For many businesses, customers are expensive to acquire, which is why losing them through churn is very costly. This why it is important for companies to capitalize on those acquisition costs by minimizing churn.

There are various options for choosing an activation function, and the choice often depends on the particular use case or application. The three most important activation functions used in state-of-the-art artificial neural networks are the (1) sigmoid or logistic activation function, (2) rectification linear unit or ReLU, and (3) softmax activation function. The sigmoid function is typically used for the neurons in the output layer since their outputs can be interpreted probabilistically.

The most common cost functions used in artificial neural networks are called (1) quadratic cost, (2) cross-entropy cost, (3) exponential cost, (4) Hellinger distance, (5) Kullback-Leibler divergence, (6) generalized Kullback-Leibler divergence, and (7) Itakura-Saito distance. The particular choice depends on the particular use case and network architecture.

The first working but not very efficient training algorithm for multilayered artificial neural networks was in fact proposed by the Russian mathematician Alexey Ivakhnenko in 1865 [41].

This property is technically referred to as spatial invariance, which is very important for increasing the reliability of convolutional neural networks.

Their performance was mainly limited due to the so-called vanishing gradient problem, which refers to a cost function that approaches a local rather than a global minimum. The activation functions used in recurrent neural networks turned out to be particularly susceptible to this problem.

Game theory has various use cases in economics and industry as it can be used to optimize decision-making processes, such as the auction of the British 3G telecom licenses in 2000 [59]. One of its pioneers, the American mathematician John Nash, who discovered the famous Nash equilibrium, was portrayed in the Oscar winning movie A Beautiful Mind in 2001.

A nice YouTube video that visualizes the style transfer capabilities of generative adversarial networks can be found on https://youtu.be/0ueRYinz8Tk/.

See www.cambridgeconsultants.com/press-releases/turning-our-sketches-art-machine-learning/.

See https://autodeskresearch.com/projects/dreamcatcher/ for further information.

A Markov process is a probabilistic model that describes a time sequence of possible actions, in which the likelihood (or value) of an action depends on the state attained in the previous action. It is named after the Russian mathematician Andrey Andreyevich Markov.

Most Go boards either use a 9 × 9, 13 × 13, or 19 × 19 lined grid.

This term refers to an approach, in which an algorithm calculates all possible solutions and maps out the breadth of possible next moves in a decision tree in order to select the best one.

Please also see https://youtu.be/WXuK6gekU1Y/ for an insightful documentary about AlphaGo.

Formally, the American socialist Charles Tilly has in fact used this term in his working paper surveying “The old new social history and the new old social history” in 1980 for the first time but in a different context.

This is why some startup companies, like Cerebras Systems, Graphcore, and Mythic AI, do not focus on traditional von Neumann–like microchip architectures anymore and have been developing on-chip compute-and-memory designs instead.

The world’s first true GPU was the “GeForce 256,” which was released by the American semiconductor company Nvidia in 1999 without realizing its potential for parallel computing and machine learning at that time.

See www.mlperf.org/.

See www.image-net.org/ for further information.

For comparison, the WSE Gen2 is reported to be 46,225 mm² in size, while the Nvidia A100 is 826 mm² in size and houses 54.2 billion transistors. The WSE Gen2 is manufactured by TSMC, a Taiwanese multinational semiconductor contract manufacturing and design company. Please see www.cerebras.net/ for further details.

4K refers to images with a horizontal resolution of approximately 4,000 pixels.

Sensor fusion refers to the aggregation of sensory data from different disparate sources of information. The resulting information is more complete and accurate than it would be when these sources were used individually.

See www.vicarious.com/ for further information.

See www.kindred.ai/autograsp/ for further information.

See, for example, www.edn.com/autonomous-vehicles-the-electronics-road-to-making-them-safe/ for further details about the employed electronics.

ImageNet organizes an international competition for machine learning scientists every year to demonstrate the performance of algorithms by labeling more than 15 million high-resolution images with more than 22,000 classes.

Redundancy as well as the consolidation of different sources of information is particularly important for autonomous driving based on artificial neural networks in order to guarantee a certain level of reliability and safety. Researchers at the German Max Planck Institute for Intelligent Systems, for instance, recently demonstrated that small patterns in the background of the camera image to be analyzed by an artificial neural network can create so much noise that a reliable image analysis fails disastrously [94].

The correct behavior can be derived from the behavior of the driver, for example, who interprets visional information about different traffic situations and reacts appropriately by, for example, steering, breaking, and accelerating the vehicle. In other words, the driver does essentially label the data while driving the vehicle prudently.

Please see www.tesla.com/autopilotAI for further information.

See www.nvidia.com/en-us/self-driving-cars/drive-platform/hardware/.

For the most part, chatbots are typically focused on businesses, such as customer support of a company. Virtual assistants, on the other side, are developed for basically everyone and help us with our daily activities.

The power usage effectiveness is a quantitative measure for the energy efficiency and defined as the ratio between the total building energy usage and the IT energy usage.

The deoxyribonucleic acid or DNA is a macromolecule that carries our genetic makeup. It is composed of two complementary chains that coil around each other to form the famous double helix structure. It was discovered by the two molecular biologists and later Nobel laureates Francis Crick and James Watson at the Cavendish Laboratory within the University of Cambridge in 1953 based on the analysis of X-ray diffraction data acquired by the British scientist Raymond Gosling at King’s College London.

As of October 27, 2020, data taken from https://coronavirus.jhu.edu/map.html/.

Aladdin is the acronym for “Asset, Liability, Debt, and Derivative Investment Network.”

Title: Artificial Intelligence
Author: Volker Lang
Publisher: Apress
Book: Digital Fluency
Print ISBN: 978-1-4842-6773-8

Electronic ISBN: 978-1-4842-6774-5

Copyright Year: 2021
DOI: https://doi.org/10.1007/978-1-4842-6774-5_4

Springer Professional

Abstract

Please log in to get access to your license.

Dont have a licence yet? Then find out more about our products and how to get one now:

Springer Professional "Wirtschaft+Technik"

Springer Professional "Technik"

Springer Professional "Wirtschaft"

Premium Partner