Search Results

The Quest for Intelligent Machines

Arlindo Oliveira

in The Digital Mind: How Science is Redefining Humanity

This chapter addresses the question of whether a computer can become intelligent and how to test for that possibility. It introduces the idea of the Turing test, a test developed to determine, in an ... More

This chapter addresses the question of whether a computer can become intelligent and how to test for that possibility. It introduces the idea of the Turing test, a test developed to determine, in an unbiased way, whether a program running in a computer is, or is not, intelligent. The development of artificial intelligence led, in time, to many applications of computers that are not possible using “non-intelligent” programs. One important area in artificial intelligence is machine learning, the technology that makes possible that computers learn, from existing data, in ways similar to the ways humans learn. A number of approach to perform machine learning is addressed in this chapter, including neural networks, decision trees and Bayesian learning. The chapter concludes by arguing that the brain is, in reality, a very sophisticated statistical machine aimed at improving the chances of survival of its owner.Less

Interactive Learning with Mutual Explanations in Relational Domains

Ute Schmid

in Human-Like Machine Intelligence

With the growing number of applications of machine learning in complex real-world domains machine learning research has to meet new requirements to deal with the imperfections of real world data and ... More

With the growing number of applications of machine learning in complex real-world domains machine learning research has to meet new requirements to deal with the imperfections of real world data and the legal as well as ethical obligations to make classifier decisions transparent and comprehensible. In this contribution, arguments for interpretable and interactive approaches to machine learning are presented. It is argued that visual explanations are often not expressive enough to grasp critical information which relies on relations between different aspects or sub-concepts. Consequently, inductive logic programming (ILP) and the generation of verbal explanations from Prolog rules is advocated. Interactive learning in the context of ILP is illustrated with the Dare2Del system which helps users to manage their digital clutter. It is shown that verbal explanations overcome the explanatory one-way street from AI system to user. Interactive learning with mutual explanations allows the learning system to take into account not only class corrections but also corrections of explanations to guide learning. We propose mutual explanations as a building-block for human-like computing and an important ingredient for human AI partnership.Less

An Overview of the RELIEF Algorithm and Advancements

Alexandre Todorov

in Statistical Approaches to Gene X Environment Interactions for Complex Phenotypes

The aim of the RELIEF algorithm is to filter out features (e.g., genes, environmental factors) that are relevant to a trait of interest, starting from a set of that may include thousands of ... More

The aim of the RELIEF algorithm is to filter out features (e.g., genes, environmental factors) that are relevant to a trait of interest, starting from a set of that may include thousands of irrelevant features. Though widely used in many fields, its application to the study of gene-environment interaction studies has been limited thus far. We provide here an overview of this machine learning algorithm and some of its variants. Using simulated data, we then compare of the performance of RELIEF to that of logistic regression for screening for gene-environment interactions in SNP data. Even though performance degrades in larger sets of markers, RELIEF remains a competitive alternative to logistic regression, and shows clear promise as a tool for the study of gene-environment interactions. Areas for further improvements of the algorithm are then suggested.Less

Logic-based Robotics

Claude Sammut, Reza Farid, Handy Wicaksono, and Timothy Wiley

in Human-Like Machine Intelligence

This chapter explores methods for combining symbolic and sub-symbolic reasoning and learning systems to take advantage of the strengths of each approach in challenging tasks in robotics. In ... More

This chapter explores methods for combining symbolic and sub-symbolic reasoning and learning systems to take advantage of the strengths of each approach in challenging tasks in robotics. In perception, Inductive Logic Programming (ILP) can be used to learn descriptions of classes of objects and to find relations between objects. Examples are given of perception for robots in urban search and rescue. We also describe systems for learning plans and behaviours for robots. Relational learning is used to acquire abstract model of robot actions that are then used to constrain sub-symbolic learning for low-level control. Models can be variously expressed in the classical STRIPS representation or as qualitative models. A STRIPS-like model is acquired by a robot that learns to use tools and also designs new tools. A qualitative model is constructed by a robot that learns to traverse uneven terrain in urban search and rescue. The model is refined by reinforcement learning.Less

Human-like Computer Vision

Stephen Muggleton and Wang-Zhou Dai

in Human-Like Machine Intelligence

Statistical machine learning is widely used in image classification and typically 1) requires many images to achieve high accuracy and 2) does not provide support for reasoning below the level of ... More

Statistical machine learning is widely used in image classification and typically 1) requires many images to achieve high accuracy and 2) does not provide support for reasoning below the level of classification.  By contrast this paper describes an approach called machine learning approach called Logical Vision (LV) which uses a) background knowledge such as light reflection that can itself be learned and used for resolving visual ambiguities, which cannot be easily modeled using statistical approaches, b) a wider class of background models representing classical 2D shapes such as circles and ellipses, c) primitive-level statistical estimators to handle noise in real images, Our results indicate that in real images the new noise-robust version of LV using a single example (ie one-shot LV) converges to an accuracy at least comparable to thirty-shot statistical machine learner on the prediction of hidden light sources.Less

Human–Machine Collaboration for Democratizing Data Science

CléMent Gautrais, Yann Dauxais, Stefano Teso, Samuel Kolb, Gust Verbruggen, and Luc De Raedt

in Human-Like Machine Intelligence

Everybody wants to analyse their data, but only few posses the data science expertise to do this. Motivated by this observation, we introduce a novel framework and system VisualSynth for ... More

Everybody wants to analyse their data, but only few posses the data science expertise to do this. Motivated by this observation, we introduce a novel framework and system VisualSynth for human-machine collaboration in data science. Its aim is to democratize data science by allowing users to interact with standard spreadsheet software in order to perform and automate various data analysis tasks ranging from data wrangling, data selection, clustering, constraint learning, predictive modeling and auto-completion. VisualSynth relies on the user providing colored sketches, i.e., coloring parts of the spreadsheet, to partially specify data science tasks, which are then determined and executed using artificial intelligence techniques.Less

We Want Robots to See and Understand the World

Antonio Torralba and Adolfo Plasencia

in Is the Universe a Hologram?: Scientists Answer the Most Provocative Questions

Antonio Torralba, member of MIT CSAIL, opens the dialogue by describing the research he performs in the field of computer vision and related artificial intelligence (AI). He also compares the ... More

Antonio Torralba, member of MIT CSAIL, opens the dialogue by describing the research he performs in the field of computer vision and related artificial intelligence (AI). He also compares the conceptual differences and the context of the early days of artificial intelligence—where hardly any image recording devices existed—with the present situation, in which an enormous amount of data is available. Next, through the use of examples, he talks about the huge complexity faced by research in computer vision to get computers and machines to understand the meanings of what they “see” in the scenes, and the objects they contain, by means of digital cameras. As he explains afterward, the challenge of this complexity for computer vision processing is particularly noticeable in settings involving robots, or driverless cars, where it makes no sense to develop vision systems that can see if they cannot learn. Later he argues why today’s computer systems have to learn “to see” because if there is no learning process, for example machine learning, they will never be able to make autonomous decisions.Less

Human–Machine Scientific Discovery

Alireza Tamaddoni-Nezhad, David Bohan, Ghazal Afroozi Milani, Alan Raybould, and Stephen Muggleton

in Human-Like Machine Intelligence

Humanity is facing existential, societal challenges related to food security, ecosystem conservation, antimicrobial resistance, etc, and Artificial Intelligence (AI) is already playing an important ... More

Humanity is facing existential, societal challenges related to food security, ecosystem conservation, antimicrobial resistance, etc, and Artificial Intelligence (AI) is already playing an important role in tackling these new challenges. Most current AI approaches are limited when it comes to ‘knowledge transfer’ with humans, i.e. it is difficult to incorporate existing human knowledge and also the output knowledge is not human comprehensible. In this chapter we demonstrate how a combination of comprehensible machine learning, text-mining and domain knowledge could enhance human-machine collaboration for the purpose of automated scientific discovery where humans and computers jointly develop and evaluate scientific theories. As a case study, we describe a combination of logic-based machine learning (which included human-encoded ecological background knowledge) and text-mining from scientific publications (to verify machine-learned hypotheses) for the purpose of automated discovery of ecological interaction networks (food-webs) to detect change in agricultural ecosystems using the Farm Scale Evaluations (FSEs) of genetically modified herbicide-tolerant (GMHT) crops dataset. The results included novel food-web hypotheses, some confirmed by subsequent experimental studies (e.g. DNA analysis) and published in scientific journals. These machine-leaned food-webs were also used as the basis of a recent study revealing resilience of agro-ecosystems to changes in farming management using GMHT crops.Less

Motif Discovery in Protein Structure Databases

Janice Glasgow and Evan Steeg

in Pattern Discovery in Biomolecular Data: Tools, Techniques, and Applications

The field of knowledge discovery is concerned with the theory and processes involved in the representation and extraction of patterns or motifs from large databases. Discovered patterns can be used ... More

The field of knowledge discovery is concerned with the theory and processes involved in the representation and extraction of patterns or motifs from large databases. Discovered patterns can be used to group data into meaningful classes, to summarize data, or to reveal deviant entries. Motifs stored in a database can be brought to bear on difficult instances of structure prediction or determination from X-ray crystallography or nuclear magnetic resonance (NMR) experiments. Automated discovery techniques are central to understanding and analyzing the rapidly expanding repositories of protein sequence and structure data. This chapter deals with the discovery of protein structure motifs. A motif is an abstraction over a set of recurring patterns observed in a dataset; it captures the essential features shared by a set of similar or related objects. In many domains, such as computer vision and speech recognition, there exist special regularities that permit such motif abstraction. In the protein science domain, the regularities derive from evolutionary and biophysical constraints on amino acid sequences and structures. The identification of a known pattern in a new protein sequence or structure permits the immediate retrieval and application of knowledge obtained from the analysis of other proteins. The discovery and manipulation of motifs—in DNA, RNA, and protein sequences and structures—is thus an important component of computational molecular biology and genome informatics. In particular, identifying protein structure classifications at varying levels of abstraction allows us to organize and increase our understanding of the rapidly growing protein structure datasets. Discovered motifs are also useful for improving the efficiency and effectiveness of X-ray crystallographic studies of proteins, for drug design, for understanding protein evolution, and ultimately for predicting the structure of proteins from sequence data. Motifs may be designed by hand, based on expert knowledge. For example, the Chou-Fasman protein secondary structure prediction program (Chou and Fasman, 1978), which dominated the field for many years, depended on the recognition of predefined, user-encoded sequence motifs for α-helices and β-sheets. Several hundred sequence motifs have been cataloged in PROSITE (Bairoch, 1992); the identification of one of these motifs in a novel protein often allows for immediate function interpretation. Less

Defining Project Objectives

Kai R. Larsen and Daniel S. Becker

in Automated Machine Learning for Business

This section covers the first steps of a the Machine Learning Life Cycle Model; how to specify a business problem, acquire subject matter expertise, define prediction target, define unit of analysis, ... More

This section covers the first steps of a the Machine Learning Life Cycle Model; how to specify a business problem, acquire subject matter expertise, define prediction target, define unit of analysis, identify success criteria, evaluate risks, and finally, decide whether to continue a project. Focus is on who will use the model, whether management is supportive, whether the drivers of the model can be visualized, and how much value a model can produce.Less

Data-Sense and Non-Sense

Sun-ha Hong

in Technologies of Speculation: The Limits of Knowledge in a Data-Driven Society

Data-driven knowledge is increasingly produced without regard to human cognition and sensibility, yet this conflagration of machinic non-sense also demands that we adapt to its rationality. ... More

Data-driven knowledge is increasingly produced without regard to human cognition and sensibility, yet this conflagration of machinic non-sense also demands that we adapt to its rationality. Data-sense describes the popular expectation that a posthuman future is inevitable, compelling human subjects to orient their personal lives and truths in ways most compatible with machine learning and other regimes of data-driven analysis. The pursuit of the human nonconscious as a source of objective truth erodes the ground beneath the feet of the good liberal subject and concomitant ideals of agency, freedom, and self-determination.Less

Learning from the Impossible

Andrea Moro

in Impossible Languages

Understanding the nature and the structure of human language coincides with capturing the constraints which make a conceivable language possible or, equivalently, whether there are impossible ... More

Understanding the nature and the structure of human language coincides with capturing the constraints which make a conceivable language possible or, equivalently, whether there are impossible languages at all. The chapter focuses on syntax, the capacity to generate potentially infinite sentences from a fixed limited set of words: ever since Descartes this has been considered the fingerprint of human mind. Modern syntax has allowed a mathematical approach to this domain.Less

Game Analytics for Games User Research

Anders Drachen and Shawn Connor

in Games User Research

Game Analytics (GA) provides new ways to conduct user research, counteracting some of the weaknesses of traditional approaches while retaining essential compatibility with the methodologies of GUR. ... More

Game Analytics (GA) provides new ways to conduct user research, counteracting some of the weaknesses of traditional approaches while retaining essential compatibility with the methodologies of GUR. This chapter provides an overview of what GA is and how it fits within the daily operations of game development across studio sizes, with an emphasis on the intersection with GUR and the synergies that can be leveraged across analytics and user research.Less

Conclusion

David M. Day and Margit Wiesner

in Criminal Trajectories: A Developmental Perspective

It has been 25 years since the criminal trajectory methodology was first introduced. Scientists from multiple fields have now arrived at a much more balanced view of its strengths and weaknesses. The ... More

It has been 25 years since the criminal trajectory methodology was first introduced. Scientists from multiple fields have now arrived at a much more balanced view of its strengths and weaknesses. The final chapter of this book looks back at the accumulated research on criminal trajectories and renews the call on criminological trajectory researchers to interface better with contemporary developmental science frameworks. This call is not intended to replace extant developmental and life-course theories of crime but, rather, to complement them by incorporating meta-theoretical propositions from the field of developmental science. To this end, this chapter offers 12 suggestions for the next generation of trajectory researchers. They range from methodological issues, including the need for stricter reporting standards and greater methodological rigor, to substantive research needs, such as the exploration of the role of biological processes, and the study of prospective links to trajectory groups of distinct behaviors and intentional self-regulatory strategies that foster desisting pathways of crime.Less

Remembering Our Future: The Frontier of Search Technologies

Ricardo Baeza-Yates and Adolfo Ricardo

in Is the Universe a Hologram?: Scientists Answer the Most Provocative Questions

In this dialogue, the computer scientist Ricardo Baeza-Yates explains why search technologies enable behavioral patterns which are more deterministic for some people rather than others to be ... More

In this dialogue, the computer scientist Ricardo Baeza-Yates explains why search technologies enable behavioral patterns which are more deterministic for some people rather than others to be predicted. And why people have in their behavior a long tail. He explains later why we feel more comfortable with determinism, which is one reason for the adoption of the semantic web taking so long. Ricardo doesn’t believe in the existence of a Kibernos (kubernetes) steering the Internet. He argues that the fact the Internet has not been dominated by any power in the physical world is not related to its size, but rather to its diversity. He then explains that artificial intelligence is related both to a complex critical mass and to computing capabilities. He also argues why not everyone is able to innovate— there is no algorithm that exists to achieve success— and why innovation is something that cannot be replicated. Finally, he explains how the Internet is broadening our social horizons and is a mirror of what we are; the way it allows us to control the devil in us, and how their mechanisms exercise control over the misuse of technology.Less

Statistical Tests and Algorithms

Áki J. Láruson and Floyd A. Reed

in Population Genetics with R: An Introduction for Life Scientists

In this chapter statistical tests are performed in R, including the χ‎2 tests; testing can be started if allele and genotype frequency data match the reader’s expectations. Writing and running ... More

In this chapter statistical tests are performed in R, including the χ‎2 tests; testing can be started if allele and genotype frequency data match the reader’s expectations. Writing and running algorithms are introduced as functions, specifically working through building an Expectation Maximization function. The function is then used to estimate and update the reader’s inferences of allele frequencies from observed and expected genotypes.Less

Who profits?

Harold Thimbleby

in Fix IT: See and solve the problems of digital healthcare

Who is profiting from our data? Is Artificial Intelligence (AI) the solution to better healthcare?

Punching above your weight: how small studios can leverage data for an unfair advantage

Lysiane Charest

in Games User Research

This chapter is aimed at small-to-medium-sized studios wanting to introduce analytics into their development process. It focuses on concepts and techniques that are most useful for smaller studios, ... More

This chapter is aimed at small-to-medium-sized studios wanting to introduce analytics into their development process. It focuses on concepts and techniques that are most useful for smaller studios, and that require minimal skills. While money is always an issue, plenty of free analytics tools exist, whether they are third-party tools or simple in-house solutions. The chapter details how the most important factor is the availability of human resources.Less