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. For example, in modern automated manufacturing facilities, there is a need to predict machine failures before they occur by analyzing sensor readings. Because the machines are new, there are no human experts who can be interviewed by a programmer to provide the knowledge necessary to build a computer system. A machine learning system can study recorded data and subsequent machine failures and learn prediction rules. Second, there are problems where human experts exist, but where they are unable to explain their expertise. This is the case in many perceptual tasks, such as speech recognition, hand-writing recognition, and natural language understanding. Virtually all humans exhibit expert-level abilities on these tasks, but none of them can describe the detailed steps that they follow as they perform them. Fortunately, humans can provide machines with examples of the inputs and correct outputs for these tasks, so machine learning algorithms can learn to map the inputs to the outputs. Third, there are problems where phenomena are changing rapidly. In finance, for example, people would like to predict the future behavior of the stock market, of consumer purchases, or of exchange rates. These behaviors change frequently, so that even if a programmer could construct a good predictive computer program, it would need to be rewritten frequently. A learning program can relieve the programmer of this burden by constantly modifying and tuning a set of learned prediction rules. Fourth, there are applications that need to be customized for each computer user separately. Consider, for example, a program to filter unwanted electronic mail messages. Different users will need different filters. It is unreasonable to expect each user to program his or her own rules, and it is infeasible to provide every user with a software engineer to keep the rules up-to-date. A machine learning system can learn which mail messages the user rejects and maintain the filtering rules automatically. Machine learning addresses many of the same research questions as the fields of statistics, data mining, and psychology, but with differences of emphasis. Statistics focuses on understanding the phenomena that have generated the data, often with the goal of testing different hypotheses about those phenomena. Data mining seeks to find patterns in the data that are understandable by people. Psychological studies of human learning aspire to understand the mechanisms underlying the various learning behaviors exhibited by people (concept learning, skill acquisition, strategy change, etc.).

Machine learning

The computational brain: Patricia S. Churchland and Terrence J. Sejnowski (MIT Press, Cambridge, MA, 1992); xi, 544 pages, $39.95

Dimensions of Reinforcement Learning

Artificial Intelligence and Tutoring Systems

The Brain's Sense of Movement. By Alain Berthoz (Translated by Giselle Weiss). Cambridge, Massachusetts: Harvard University Press; 2000, 352 pp. $22.80. Ever wonder how certain people catch or bat a baseball hurled at blurring speeds? If you have, find yourself in a group whose intended or accidental success maybe a machine that pitches and throws like a ballplayer. Once this group of researchers articulates an accurate set of principles behind movement, deft engineering, persistence, and luck may converge to emulate nature. Although Berthoz's The Brain's Sense ofMovement, does not offer a science-fiction glimpse of agile androids that populate Asimov's novels, it provides an organized and fascinating way of thinking about movement. Berthoz takes the reader on a whirlwind tour of cognitive neuroscience topics: perception, coherence, memory, prediction, and adaptation. By examining these topics and using choice examples, Berthoz builds a persuasive case supporting his thesis that the brain is an anticipation machine. Even before delving into the intricacies of each of these topics, Berthoz's claim seems reasonable in light of evolution. In fact, Berthoz explains how evolution and improved neural systems that guide movement influence and drive each other: \"The species that passed the test of natural selection are those that figured out how to save a few milliseconds in capturing prey and anticipating the actions of predators, those whose brains were able to simulate the elements of the environment and choose the best way home, those able to memorize great quantities of information from past experience and use them in the heat of action.\" This cat and mouse games has honed the brain to take advantage of its parallel architecture, bypassing computing each trajectory in a Newtonian sense, and arriving at a solution by using heuristics developed over evolution. Heuristics play an important role in examples where a target exceeds physical limits of detection. For example, a baseball may move too quickly for the fovea to focus, however, the brain, and skeletal-muscular system use computational shortcuts to simulate, predict, adapt, and control the body in response to a changing environment. The first choice example that Berthoz highlights as a key computational shortcut is the derivative. Signals from receptors enable anticipation of future position of the head owing to their sensitivity to derivatives such as jerk, acceleration, and velocity. Another mathematical concept that Berthoz explores as a predictive tool is tensors. From what I learned, a tensor is a group of mathematical operators called matrices that carry out transformations among vectors. Between derivatives and vectors, Berthoz devotes several chapters to explaining how otoliths and semicircular canals use derivatives for linear and angular accelerations to predict while tensors receive perfunctory treatment. A balance between these two topics may better satisfy some readers. Certainly, derivatives and tensors alone cannot account for movement. Just as a calculator or computer derives its usefulness in a network, mathematical shortcuts for movement need to occur in the context of a circuit. Reading Berthoz's

The Brain's Sense of Movement.

Many Virtual Reality (VR) applications, such as Virtual Learning Environments or Interactive Virtual Tours, are based on a rich semantic description of the environment and tasks that users have to perform. These applications are built upon Virtual Environments (VEs) in which artificial agents act autonomously while interacting in realtime with users. Semantic modelling of a VR environment makes it possible the knowledge-driven access from the description of VEs that simplifies the development of VR applications. It eases the development of these types of applications. Semantic modelling should provide a consistent representation of the following aspects: 1) The simulated world, its structure and the behavior of its entities, 2) Interactions and tasks, that users and agents can perform in the environment, 3) Knowledge items, that autonomous agents can use for decision-making or for communication with users. This paper presents MASCARET, a model-based approach, for the design of semantic VR environments. This approach is based on the Unified Modeling Language (UML). In this approach, UML is used to provide a knowledge-driven access to the semantic contents of the VE and not for code generation, as in classical software development process. Interests of a UML-based approach are that its metamodel covers different views of the semantic modelling: ontology, structure, behaviors, interactions, activities. It is also an extensible language that can be specialized to provide formal operational semantics. We also present how MASCARET can be used to develop content-rich interactive applications that can be deployed over various VR platforms. Finally, we discuss the benefits of such a metamodel-based approach and show how the multi-layer semantic model can be used in different VR applications, in which adaptive behaviors of artificial agents acting within complex environments have to be simulated.

Semantic modeling of Virtual Environments using MASCARET

HAL is a multi-disciplinary open access archive for the deposit and dissemination of scientific research documents, whether they are published or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers. L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés. Mascaret: Pedagogical multi-agents system for virtual environment for training. Cédric Buche, Ronan Querrec, Pierre De Loor, Pierre Chevaillier

MASCARET: A Pedagogical Multi-Agent System for Virtual Environments for Training

This article deals with the links between the enaction paradigm and artificial intelligence. Enaction is considered a metaphor for artificial intelligence, as a number of the notions which it deals with are deemed incompatible with the phenomenal field of the virtual. After explaining this stance, we shall review previous works regarding this issue in terms of artificial life and robotics. We shall focus on the lack of recognition of co-evolution at the heart of these approaches. We propose to explicitly integrate the evolution of the environment into our approach in order to refine the ontogenesis of the artificial system, and to compare it with the enaction paradigm. The growing complexity of the ontogenetic mechanisms to be activated can therefore be compensated by an interactive guidance system emanating from the environment. This proposition does not however, resolve that of the relevance of the meaning created by the machine (sense-making). Such reflections lead us to integrate human interaction into this environment in order to construct relevant meaning in terms of participative artificial intelligence. This raises a number of questions with regards to setting up an enactive interaction. The article concludes by exploring a number of issues, thereby enabling us to associate current approaches with the principles of morphogenesis, guidance, the phenomenology of interactions and the use of minimal enactive interfaces in setting up experiments which will deal with the problem of artificial intelligence in a variety of enaction-based ways.

https://hal.archives-ouvertes.fr/hal-00654120/document

Enaction-Based Artificial Intelligence: Toward Co-evolution with Humans in the Loop

In this paper, we address the problem of creating believable agents (virtual characters) in video games. We consider only one meaning of believability, ``giving the feeling of being controlled by a player'', and outline the problem of its evaluation. We present several models for agents in games which can produce believable behaviours, both from industry and research. For high level of believability, learning and especially imitation learning seems to be the way to go. We make a quick overview of different approaches to make video games' agents learn from players. To conclude we propose a two-step method to develop new models for believable agents. First we must find the criteria for believability for our application and define an evaluation method. Then the model and the learning algorithm can be designed.

https://www.enib.fr/~buche/article/GAMEON_10.pdf

The challenge of believability in video games: Definitions, agents models and imitation learning

Spatial objects and relationships between them, compose a spatial model that is the backbone of virtual environments (VEs). However, due to the natural complexity of both spatial objects and spatial information, the modeling of such spatial relationships is still a difficult task. This paper presents a novel approach for representing semantic spatial relations in VEs using the Unified Modeling Language (UML) and the Object Constraint Language (OCL). Our approach first uses the UML class model as a conceptual model for VEs. We then propose a spatial extension of OCL named VRX-OCL as a high-level and flexible language to cover multidimensional, manifold, and reference frame-dependent spatial constraints. We mainly focus on two important classes of spatial relations, namely, topological and projective relations that allow nonmetric representation of space. The applicability of our approach is demonstrated in the Virtual Physics Laboratory, a VE for learning physics. Based on the constraints satisfaction, the system is able to visualize abstract spatial information and thus provides educational assistance to the learners.

/pdf/ensuring-semantic-spatial-constraints-in-virtual-vljufaj7pl.pdf

Pierre De Loor

Papers

Semantic modeling of Virtual Environments using MASCARET

MASCARET: A Pedagogical Multi-Agent System for Virtual Environments for Training

Enaction-Based Artificial Intelligence: Toward Co-evolution with Humans in the Loop

The challenge of believability in video games: Definitions, agents models and imitation learning

Ensuring semantic spatial constraints in virtual environments using UML/OCL