By Topic

Machine Learning, IEE Colloquium on

Date 28 Jun 1990

Filter Results

Displaying Results 1 - 11 of 11
  • Neural learning algorithms: some empirical trials

    Page(s): 8/1 - 8/7
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (304 KB)  

    Reports some empirical trials comparing seven different neural learning algorithms (including two versions of back propagation) on four test problems. Though limited in scope the study does shed light on the performance of a variety of learning techniques, compared under relatively uniform conditions. The results cast some doubt on the status of back propagation as an `industrial strength' learning algorithm View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Machines can learn actions but can they classify situations?

    Page(s): 4/1 - 4/3
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (172 KB)  

    The use of the genetic algorithm for learning individual actions in a rule-based process control system is described. Its use for learning a whole set of actions for controlling a dynamical system is then described with two possible classifications of the state space. How the genetic algorithm can be used to learn classifications in a database application is then outlined and the relevance of this to state-space classification shown. Finally, the difficulties of combining the different processes is discussed View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • IEE Colloquium on `Machine Learning' (Digest No.117)

    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (12 KB)  

    The following topics were dealt with: machine learning applications in: civil engineering project management, n-Tuple network mapping optimisation; situation classification, handprinted character recognition, natural language interfaces, planning, and electronic filter tuning; example set quality in induction; neutral learning algorithms; and hybrid genetic algorithms View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Hybrid genetic algorithms for machine learning

    Page(s): 9/1 - 9/3
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (128 KB)  

    Describes the basic genetic algorithm and then discuss some ways in which it may be hybridized with other types of optimisation techniques. The comments on hybridization are of two kinds. First, three general principles for hybridizing genetic algorithms and other algorithms are given. These principles have often generated hybrid optimization systems that perform better than the systems they arose from. Second, two examples of such hybrid systems, one of optimizing the design of packet-switching telecommunications networks, and one of training neural networks, are described View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Empirical results from applying machine learning techniques to planning

    Page(s): 6/1 - 6/3
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (192 KB)  

    Outlines an experimental machine learning implementation, called `FM', that applies both explanation based learning and similarity-based learning to AI planners. The system shell of FM contains techniques for learning application-dependent heuristics, through the experience of using a performance component (a planner) in that application. An application domain is supplied by specifying a set of action schemas, and environmental facts and rules. FM is then fed an initial state, and a sequence of tasks within this application, roughly in ascending order of complexity, which it is expected to solve. After each task has been solved, the system analyses the planning trace, allowing it to learn from experience View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Machine learning for handprinted character perception

    Page(s): 5/1 - 5/5
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (228 KB)  

    Presents a functional model of machine learning by means of inductive inference to read handprinted text. This model can be used to characterise, compare and act as a generic model for methods of machine learning. The environment determines the design of the learning system. The proposed model of machine learning and perception of handwritten text has achieved character error rates comparable to those commonly experienced by humans when reading isolated handprinted characters View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Learning by analytical methods or induction: a case study

    Page(s): 10/1 - 10/3
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (172 KB)  

    An expert system is under construction for the application of tuning electronic filters. Machine learning techniques were used to enhance knowledge elicitation and experiences gained in implementing them are presented in terms of learning, testing, and learning refinement View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Evolutionary learning to optimise mapping in n-Tuple networks

    Page(s): 3/1 - 3/3
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (140 KB)  

    The use of n-Tuple networks as pattern recognition devices is well known. Networks are able to recognise and discriminate between different classes of data if each class is taught into a separate discriminator. If the different classes are too similar, however, the system can have difficulty discriminating between the classes. This paper describes a novel method of optimising the learning strategy to minimise this problem View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • The importance of example set quality in induction

    Page(s): 2/1 - 2/3
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (124 KB)  

    Inductive algorithms start from a set of examples of a solved problem and generate rules which classify the solutions to the problem in terms of attributes describing the problem. Inductive methods are not necessarily easy to apply, and success is likely to depend on the quality of the example set in the chosen domain. Recognition before acquisition of the likelihood of obtaining a good quality example set would clearly be desirable, and some parameters leading towards such recognition have been identified View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Managing risk in civil engineering by machine learning from failures

    Page(s): 1/1 - 1/4
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (204 KB)  

    Describes the production of `a knowledge based computer system which may be an aid in the management of a project'. The newly emerging technologies of artificial intelligence and `expert systems', and in particular machine learning from examples, present new opportunities for the development of management tools View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Learning dialogue at the interface

    Page(s): 7/1 - 7/3
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (104 KB)  

    The authors introduce a knowledge based system, called GAMINT, which has been developed to act as a natural language interface to a generalised computer system. The system features two primary objectives: it affords a means to create sophisticated dialogue at the interface; and it allows incremental learning to take place View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.