If an expert system--brilliantly designed, engineered and implemented--cannot learn not to repeat its mistakes, it is not as intelligent as a worm or a sea anemone or a kitten. -Oliver G. Selfridge, from The Gardens of Learning. "Find a bug in a program, and fix it, and the program will work today. Show the program how to find and fix a bug, and the program will work forever." - Oliver G. Selfridge, in AI's Greatest Trends and Controversies

Oliver Selfridge

Machine learning refers to a system capable of the autonomous acquisition and integration of knowledge. This capacity to learn from experience, analytical observation, and other means, results in a system that can continuously self-improve and thereby offer increased efficiency and effectiveness.

Introductory Readings

The Discipline and Future of Machine Learning. Video of Tom Mitchell's March 1, 2007 seminar talk at the Carnegie Mellon University School of Computer Science's Machine Learning Department: "Over the past 50 years the study of machine learning has grown from the efforts of a handful of computer engineers exploring whether computers could learn to play games, and a field of statistics that largely ignored computational considerations, to a broad discipline that has produced fundamental statistical-computational theories of learning processes, has designed learning algorithms that are routinely used in commercial systems from speech recognition to computer vision, and has spun off an industry in data mining to discover hidden regularities in the growing volume of online data. This talk will provide a personal view of the current state of machine learning, and where I think the field might (should) be headed over the coming decade. I’ll propose several specific research areas which seem to me to have great potential, and will leave plenty of time at the end for audience discussion."

• Also ...
  • watch this Interview with Tom Mitchell from the 2006 CMU Machine Learning Autumn School video collection available from VideoLectures;
  • read Does Machine Learning Really Work? By Tom M. Mitchell. AI Magazine 18(3): Fall 1997, 11-20. "Yes. Over the past decade, machine learning has evolved from a field of laboratory demonstrations to a field of significant commercial value. ... This article, based on the keynote talk presented at the Thirteenth National Conference on Artificial Intelligence, samples a number of recent accomplishments in machine learning and looks at where the field might be headed."
  • see this interview with Tom Mitchell;
  • read his whitepaper on The Discipline of Machine Learning;
  • visit the Machine Learning Department at Carnegie Mellon University;
  • ... and check out Tom Mitchell's homepage with resources that address the questions: How can we make computers improve automatically from experience? and Where do I think the study of intelligence is headed?

Machine Learning. An AI Bite by Simon Colton. Sponsored by, and available from, The Society for the Study of Artificial Intelligence and Simulation of Behaviour. "Given a task we want the computer to do, the idea is to repeatedly demonstrate how the task is performed, and let the computer learn by example, i.e., generalise some rules about the task and turn these into a program."

Introduction to Machine Learning - Draft of Incomplete Notes. By Nils J. Nilsson. "The notes survey many of the important topics in machine learning circa 1996. My intention was to pursue a middle ground between theory and practice. The notes concentrate on the important ideas in machine learning---it is neither a handbook of practice nor a compendium of theoretical proofs. My goal was to give the reader sufficient preparation to make the extensive literature on machine learning accessible."

• Chapter 1, Introduction: What is Machine Learning? "Machine learning usually refers to the changes in systems that perform tasks associated with artificial intelligence (AI). Such tasks involve recognition, diagnosis, planning, robot control, prediction, etc. ... To be slightly more specific, we show the architecture of a typical AI 'agent' in Fig. 1.1. ... One might ask 'Why should machines have to learn? Why not design machines to perform as desired in the first place?' There are several reasons why machine learning is important. ..."

Software That Learns by Doing. Machine-learning techniques have been used to create self-improving software for decades, but recent advances are bringing these tools into the mainstream. By Gary H. Anthes. Computerworld (February 6, 2006). "Attempts to create self-improving software date to the 1960s. But 'machine learning,' as it's often called, has remained mostly the province of academic researchers, with only a few niche applications in the commercial world, such as speech recognition and credit card fraud detection. Now, researchers say, better algorithms, more powerful computers and a few clever tricks will move it further into the mainstream. ... Computer scientist Tom Mitchell, director of the Center for Automated Learning and Discovery at Carnegie Mellon University, says machine learning is useful for the kinds of tasks that humans do easily -- speech and image recognition, for example -- but that they have trouble explaining explicitly in software rules. In machine-learning applications, software is 'trained' on test cases devised and labeled by humans, scored so it knows what it got right and wrong, and then sent out to solve real-world cases. Mitchell is testing the concept of having two classes of learning algorithms in essence train each other...."

Machine learns games 'like a human.' By Will Knight. New Scientist News (January 24, 2005). "A computer that learns to play a 'scissors, paper, stone' by observing and mimicking human players could lead to machines that automatically learn how to spot an intruder or perform vital maintenance work, say UK researchers. CogVis, developed by scientists at the University of Leeds in Yorkshire, UK, teaches itself how to play the children's game by searching for patterns in video and audio of human players and then building its own 'hypotheses' about the game's rules. In contrast to older artificial intelligence (AI) programs that mimic human behaviour using hard-coded rules, CogVis takes a more human approach, learning through observation and mimicry, the researchers say. ... 'A system that can observe events in an unknown scenario, learn and participate just as a child would is almost the Holy Grail of AI,' says Derek Magee from the University of Leeds." Be sure to see the sidebar with related articles & web sites.

Machine Learning Lecture Notes. From Professor Charles R. Dyer, University of Wisconsin - Madison.

Machine Learning. Section 1.2.8 of Chapter One (available online) of George F. Luger's textbook, Artificial Intelligence: Structures and Strategies for Complex Problem Solving, 5th Edition (Addison-Wesley; 2005). "The importance of learning, however, is beyond question, particularly as this ability is one of the most important components of intelligent behavior. ... Although learning is a difficult area, there are several programs that suggest that it is not impossible. One striking program is AM, the Automated Mathematician, designed to discover mathematical laws (Lenat 1977, 1982). Initially given the concepts and axioms of set theory, AM was able to induce such important mathematical concepts as cardinality, integer arithmetic, and many of the results of number theory. AM conjectured new theorems by modifying its current knowledge base and used heuristics to pursue the 'best' of a number of possible alternative theorems. ... Early influential work includes Winston's research on the induction of structural concepts such as 'arch' from a set of examples in the blocks world (Winston 1975 a)."

Machine Learning. Preprint of Thomas G. Dietterich's article in Nature Encyclopedia of Cognitive Science, London: Macmillan, 2003. Available from the author's collection of introductory information. "Machine Learning is the study of methods for programming computers to learn. Computers are applied to a wide range of tasks, and for most of these it is relatively easy for programmers to design and implement the necessary software. However, there are many tasks for which this is difficult or impossible. These can be divided into four general categories. First, there are problems for which there exist no human experts. ... Second, there are problems where human experts exist, but where they are unable to explain their expertise. ... Third, there are problems where phenomena are changing rapidly. ... Fourth, there are applications that need to be customized for each computer user separately."

• Also see Professor Dietterich's home page for links to ML resources and more information about his research at Oregon State University: "The focus of my research is machine learning: How can we make computer systems that adapt and learn from their experience? How can we combine human knowledge with massive data sets to expand scientific knowledge and build more useful computer applications? My laboratory combines research on machine learning fundamentals with applications to problems in science and engineering."

Two courses from MIT's OpenCourseWare, "a free and open educational resource for faculty, students, and self-learners around the world."

• Artificial Intelligence; Spring 2003. Professors Tomás Lozano-Pérez & Leslie Kaelbling. See the

Videos of lectures & interviews from the 2006 Machine Learning Autumn School at CMU: Machine Learning over Text & Images (available from VideoLectures). "Machine learning approaches to natural language processing problems such as information retrieval, document classification, and information extraction have developed rapidly over recent years. Even more recently, the joint analysis of text and images has become a significant focus for machine learning. This autumn school will summarize the state of the art in machine learning for text analysis and for joint text/image analysis, as presented by researchers active in these fields. It is intended for students who already have a familiarity with machine learning, and is designed for software developers, graduate students, and advanced researchers with an interest in learning more about this area."

Glossary of Terms. Special Issue on Applications of Machine Learning and the Knowledge Discovery Process. Ron Kohavi and Foster Provost, eds. Machine Learning, 30: 271-274 (1998). "To help readers understand common terms in machine learning, statistics, and data mining, we provide a glossary of common terms."

General Readings

Applying Metrics to Machine-Learning Tools: A Knowledge Engineering Approach. Fernando Alonso, Luis Mate, Natalia Juristo, Pedro L. Munoz, and Juan Pazos. AI Magazine 15(3): Fall 1994, 63-75. "The field of knowledge engineering has been one of the most visible successes of AI to date. Knowledge acquisition is the main bottleneck in the knowledge engineer's work. Machine-learning tools have contributed positively to the process of trying to eliminate or open up this bottleneck, but how do we know whether the field is progressing? How can we determine the progress made in any of its branches? How can we be sure of an advance and take advantage of it? This article proposes a benchmark as a classificatory, comparative, and metric criterion for machine-learning tools. The benchmark centers on the knowledge engineering viewpoint, covering some of the characteristics the knowledge engineer wants to find in a machine-learning tool."

Machine Learning: A Historical and Methodological Analysis. By Jaime G. Carbonell, Ryszard S. Michalski, and Tom M. Mitchell. AI Magazine 4(3): Fall 1983, 69-79. Abstract: "Machine learning has always been an integral part of artificial intelligence, and its methodology has evolved in concert with the major concerns of the field. In response to the difficulties of encoding ever-increasing volumes of knowledge in modern AI systems, many researchers have recently turned their attention to machine learning as a means to overcome the knowledge acquisition bottleneck. This article presents a taxonomic analysis of machine learning organized primarily by learning strategies and secondarily by knowledge representation and application areas. A historical survey outlining the development of various approaches to machine learning is presented from early neural networks to present knowledge-intensive techniques."

Brain learns like a robot - Scan shows how we form opinions. By Tanguy Chouard. Nature Science Update (June 10, 2004). "Researchers may have pinpointed the brain regions that help us work out good from bad. And their results suggest that humans and robots are more alike than we may care to admit, as both use similar strategies to make value judgements. ... The team also plotted brain activity on a graph to give a mathematical description of processes that underlie the formation of value judgements. The patterns they saw resembled those made by robots as they learn from experience. 'The results were astounding,' says study co-author Peter Dayan. 'There was an almost perfect match between the brain signals and the numerical functions used in machine learning,' he says. This suggests that our brains are following the laws of artificial intelligence."

Machine Learning Research: Four Current Directions. By Tom Dietterich. AI Magazine 18(4): Winter 1997, 97-136. Abstract: "Machine-learning research has been making great progress in many directions. This article summarizes four of these directions and discusses some current open problems. The four directions are (1) the improvement of classification accuracy by learning ensembles of classifiers, (2) methods for scaling up supervised learning algorithms, (3) reinforcement learning, and (4) the learning of complex stochastic models."

Learning. An overview by Patrick Doyle. Very informative, though there are some spots that are quite technical.

Journal of Machine Learning Research. "The Journal of Machine Learning Research (JMLR) provides an international forum for the electronic and paper publication of high-quality scholarly articles in all areas of machine learning."

Machine Learning, Neural and Statistical Classification. Donald Michie, D. J. Spiegelhalter, and C. C. Taylor, editors. "[This] book (originally published in 1994 by Ellis Horwood) is now out of print. The copyright now resides with the editors who have decided to make the material freely available on the web."

A video file of Tom Mitchell's AAAI Presidential Address, August 2002 is available from his home page.] "Thesis of This Talk: The synergy between AI and Brain Sciences will yield profound advances in our understanding of intelligence over the coming decade, fundamentally changing the nature of our field."

Statistical Data Mining Tutorials - Tutorial Slides by Andrew Moore, professor of Robotics and Computer Science at the School of Computer Science, Carnegie Mellon University. "The following links point to a set of tutorials on many aspects of statistical data mining, including the foundations of probability, the foundations of statistical data analysis, and most of the classic machine learning and data mining algorithms."

Machine learning on physical robots [slide show with audio]. By Peter Stone, The University of Texas at Austin. Presented at the University of Pennsylvania's GRASP Lab's GRASP Seminar Series (March 31, 2006). Abstract: "As robot technology advances, we are approaching the day when robots will be deployed prevalently in uncontrolled, unpredictable environments: the proverbial 'real world.' As this happens, it will be essential for these robots to be able to adapt autonomously to their changing environment. For a robot to learn to improve its performance based entirely on real-world environmental feedback, the robot's behavior specification and learning algorithm must be constructed so as to enable data-efficient learning. This talk presents 3 examples of machine learning on physical robots. ..."

Automated Learning and Discovery State-Of-The-Art and Research Topics in a Rapidly Growing Field. By Sebastian Thrun, Christos Faloutsos, Tom Mitchell, and Larry Wasserman. AI Magazine 20(3): Fall 1999, 78-82. "This article summarizes the Conference on Automated Learning and Discovery (CONALD), which took place in June 1998 at Carnegie Mellon University. CONALD brought together an interdisciplinary group of scientists concerned with decision making based on data. One of the meeting's focal points was the identification of promising research topics, which are discussed toward the end of this article."

Related Resources

AI on the Web: Machine Learning. A resource companion to Stuart Russell and Peter Norvig's "Artificial Intelligence: A Modern Approach" with links to reference material, people, research groups, books, companies and much more.

"The Adaptive Systems Group at the Navy Center for Applied Research in Artificial Intelligence is performing state-of-the-art research in machine learning and robotics, with emphasis on techniques that will allow the creation of systems that are more adaptive to changes in their environment and to changes in their own capabilities." Check out their research projects, such as:

• Continuous and Embedded Learning (Anytime Learning): "Continuous and embedded learning is a general approach to continuous learning in a changing environment. ... The basic idea is to integrate two continuously running modules: an execution module and a learning module. This work is part of an ongoing investigation of machine learning techniques for solving sequential decision problems."

Applications of Machine Learning collection from the Alberta Ingenuity Centre for Machine Learning.

"Grammatical Inference, variously refered to as automata induction, grammar induction, and automatic language acquisition, refers to the process of learning of grammars and languages from data. Machine learning of grammars finds a variety of applications in syntactic pattern recognition, adaptive intelligent agents, diagnosis, computational biology, systems modelling, prediction, natural language acquisition, data mining and knowledge discovery. ... This homepage is designed to be a centralized resource information on Grammatical Inference and its applications. We hope that this information will be useful to both newcomers to the field as well as seasoned campaigners"

Index of Machine Learning Courses. Maintained by Vasant Honavar, Artificial Intelligence Research Group, Department of Computer Science, Iowa State University. When you visit the page for any given course, be sure to check out sections such as 'course readings' and 'additional resources' for you're sure to find plenty of gems there.

MLnet OiS. "Welcome to the MLnet Online Information Service (the successor of the ML-Archive at GMD). This site is dedicated to the field of machine learning, knowledge discovery, case-based reasoning, knowledge acquisition, and data mining. Get information about research groups and persons within the community. Browse through the list of software and data sets, and check out our events page for the latest calls for papers. Alternatively have a look at our list of job offerings if you are looking for a new opportunity within the field." This web site is funded by the European Commission. Here are some links to just a few of their collections:

Machine Learning at IBM. "The Machine Learning Group [Haifa] specializes in developing algorithms for automatic pattern recognition, prediction, analysis, classification, and learning of structures."

Machine Learning and Applied Statistics at Microsoft. "The Machine Learning and Applied Statistics (MLAS) group is focused on learning from data and data mining. By building software that automatically learns from data, we enable applications that (1) do intelligent tasks such as handwriting recognition and natural-language processing, and (2) help human data analysts more easily explore and better understand their data."

The Machine Learning Department, an academic department within Carnegie Mellon University's School of Computer Science and successor to CALD, the Center for Automated Learning and Discovery. "We focus on research and education in all areas of statistical machine learning."

• "What is Machine Learning? Machine Learning is a scientific field addressing the question 'How can we program systems to automatically learn and to improve with experience?' We study learning from many kinds of experience, such as learning to predict which medical patients will respond to which treatments, by analyzing experience captured in databases of online medical records. We also study mobile robots that learn how to successfully navigate based on experience they gather from sensors as they roam their environment, and computer aids for scientific discovery that combine initial scientific hypotheses with new experimental data to automatically produce refined scientific hypotheses that better fit observed data. To tackle these problems we develop algorithms that discover general conjectures and knowledge from specific data and experience, based on sound statistical and computational principles. We also develop theories of learning processes that characterize the fundamental nature of the computations and experience sufficient for successful learning in machines and in humans."

Machine Learning Dictionary. Compiled by Bill Wilson, Associate Professor in the Artificial Intelligence Group, School of Computer Science and Engineering, University of NSW. "You should use The Machine Learning Dictionary to clarify or revise concepts that you have already met. The Machine Learning Dictionary is not a suitable way to begin to learn about Machine Learning."

Machine Learning in Games. Maintained by Jay Scott. "How computers can learn to get better at playing games. This site is for artificial intelligence researchers and intrepid game programmers. I describe game programs and their workings; they rely on heuristic search algorithms, neural networks, genetic algorithms, temporal differences, and other methods. I keep big list of online research papers. And there's more."

Machine Learning and Inference (MLI) Laboratory at George Mason University (GMU) "conducts fundamental and experimental research on the development of intelligent systems capable of advanced forms of learning, inference, and knowledge generation, and applies them to real-world problems."

Machine Learning Resources. Maintained by David Aha. Links to a wealth of information await you at this site.

Other References Offline

Abu-Mostafa, Yaser. 1995. Machines That Learn From Hints. Scientific American 272(4) (April 1995): 64-69. Machine learning improves significantly by taking advantage of information available from intelligent hints.

Dietterich, Thomas G. 1990. Machine Learning. In Annual Review of Computer Science, Volume 4,1989-1990, ed. Traub, Joseph F., Barbara J. Grosz, Butler W. Lampson, et al., Palo Alto, CA: Annual Reviews, Inc

Kaelbling, L. P., M. L. Littman, and A. W. Moore. 1996. Reinforcement Learning: A Survey. Journal of Artificial Intelligence Research 4: 237-285.

Kearns, M., and U. Vazirani. 1994. An Introduction to Computational Learning Theory. Cambridge, MA: MIT Press.

Langley, Pat. 1995. Elements of Machine Learning. San Francisco: Morgan Kaufmann.

Luger, George F. 2004. Artificial Intelligence: Structures and Strategies for Complex Problem Solving (5th Edition). Addison-Wesley.

Mechner, David. A. 1998. All Sytems Go. The Sciences 38 (Jan/Feb 1998): 32-7.

Michalski, Ryszard, and Georghe Tecuci, editors. 1993. Machine Learning: A Multi-Strategy Approach, Volume IV. San Francisco: Morgan Kaufmann.

Minton, Steven, editor. 1993. Machine Learning Methods for Planning. San Francisco: Morgan Kaufmann.

Mitchell, Tom. 1997. Machine Learning. McGraw-Hill.

• Slides for instructors are available.

Nilsson, Nils. 1990. The Mathematical Foundations of Learning Machines. San Francisco: Morgan Kaufmann. A reprinted version of Learning Machines: Foundations of Trainable Pattern-Classifying Systems, N. Nilsson. New York: McGraw Hill, 1965.

Patterson, Dan W. 1990. Early Work in Machine Learning. In Introduction to Artificial Intelligence and Expert Systems by Dan W. Patterson, 367-380. Englewood Cliffs, NJ: Prentice Hall.

Samuel, Arthur L. 1959. Some Studies in Machine Learning Using the Game of Checkers. In Computation and Intelligence: Collected Readings, ed. Luger, George F., Menlo Park, CA/Cambridge, MA/London: AAAI Press/The MIT Press, 1995.

Shavlik, J., and T. Dieterrich, editors. 1990. Readings in Machine Learning. San Mateo, CA: Morgan Kaufmann.

Sussman, G. 1975. A Computer Model of Skill Acquisition. Amsterdam: Elsevier/North Holland. A classic work.

Sutton, Richard S., and Andrew G. Barto. 1998. Reinforcement Learning. Cambridge, MA: MIT Press/Bradford Books. Covers the main concepts and algorithms of reinforcement learning, some history, recent developments, and applications.

Wayner, Peter. 1995. Machine Learning Grows Up. Byte 20 (August 1995): 63-64+.

Weiss, Sholom M., and Casimir A. Kulikowski. 1991. Computer Systems that Learn: Classification and Prediction Methods from Statistics, Neural Nets, Machine Learning, and Expert Systems. San Mateo, CA: Morgan Kaufmann.