An Assessment of Machine Learning Methods for Robotic Discovery

Bratko, Ivan
December 2008
Journal of Computing & Information Technology;Dec2008, Vol. 16 Issue 4, p247
Academic Journal
In this paper we consider autonomous robot discovery through experimentation in the robot's environment. We analyse the applicability of machine learning (ML) methods with respect to various levels of robot discovery tasks, from extracting simple laws among the observed variables, to discovering completely new notions that were never mentioned in the data directly. We first present some illustrative experiments in robot learning in the XPERO European project. Then we formulate a systematic list of types of learning or discovery tasks, and discuss the suitability of chosen ML methods for these tasks.


Related Articles

  • Data Preprocessing for Supervised Leaning. Kotsiantis, S. B.; Kanellopoulos, D.; Pintelas, P. E. // Enformatika;2006, Vol. 12, p277 

    Many factors affect the success of Machine Learning (ML) on a given task. The representation and quality of the instance data is first and foremost. If there is much irrelevant and redundant information present or noisy and unreliable data, then knowledge discovery during the training phase is...

  • Coordinate Descent Method for Large-scale L2-loss Linear Support Vector Machines. Kai-Wei Chang; Cho-Jui Hsieh; Chih-Jen Lin // Journal of Machine Learning Research;7/1/2008, Vol. 9 Issue 7, p1369 

    Linear support vector machines (SVM) are useful for classifying large-scale sparse data. Problems with sparse features are common in applications such as document classification and natural language processing. In this paper, we propose a novel coordinate descent algorithm for training linear...

  • User-Oriented Feature Selection for Machine Learning. Hongli Liang; Jue Wang; Yiyu Yao // Computer Journal;2007, Vol. 50 Issue 4, p421 

    The effectiveness of any machine learning algorithm depends, to a large extent, on the selection of a good subset of features or attributes. Most existing methods use the syntactic or statistical information of the data, relying on a heuristic criterion to select features. In this paper, we...

  • A review on the combination of binary classifiers in multiclass problems. Lorena, Ana Carolina; De Carvalho, André C. P. L. F.; Gama, João M. P. // Artificial Intelligence Review;Dec2008, Vol. 30 Issue 1-4, p19 

    Several real problems involve the classification of data into categories or classes. Given a data set containing data whose classes are known, Machine Learning algorithms can be employed for the induction of a classifier able to predict the class of new data from the same domain, performing the...

  • Computational Approaches to Representation Change during Learning and Development. Morrison, Clayton T.; Oates, Tim // AI Magazine;Spring2008, Vol. 29 Issue 1, p101 

    The article reports on the symposium titled "Computational Approaches to Representation Change during Learning and Development." the symposium is developed by the Association for the Advancement of Artificial Intelligence (AAAI). It aims to understand how representation change can be transferred...

  • Mathematical properties of neuronal TD-rules and differential Hebbian learning: a comparison. Kolodziejski, Christoph; Porr, Bernd; W�rg�tter, Florentin // Biological Cybernetics;Mar2008, Vol. 98 Issue 3, p259 

    A confusingly wide variety of temporally asymmetric learning rules exists related to reinforcement learning and/or to spike-timing dependent plasticity, many of which look exceedingly similar, while displaying strongly different behavior. These rules often find their use in control tasks, for...

  • Enclosing machine learning: concepts and algorithms. Xun-Kai Wei; Ying-Hong Li; Yu-Fei Li; Dong-Fang Zhang // Neural Computing & Applications;2008, Vol. 17 Issue 3, p237 

    A novel machine learning paradigm, i.e., enclosing machine learning based on regular geometric shapes was proposed. First, it adopted regular minimum volume enclosing and bounding geometric shapes (sphere, ellipsoid, box) or their unions and so on to obtain one class description model. Second,...

  • On boosting kernel density methods for multivariate data: density estimation and classification. Marzio, Marco; Taylor, Charles // Statistical Methods & Applications;2005, Vol. 14 Issue 2, p163 

    Statistical learning is emerging as a promising field where a number of algorithms from machine learning are interpreted as statistical methods and vice-versa. Due to good practical performance, boosting is one of the most studied machine learning techniques. We propose algorithms for...

  • Three New MDL-Based Pruning Techniques for Robust Rule Induction. Pham, D. T.; Afify, A. A. // Proceedings of the Institution of Mechanical Engineers -- Part C;Apr2006, Vol. 220 Issue 4, p553 

    Overfitting the training data is a major problem in machine learning, particularly when noise is present. Overfitting increases learning time and reduces both the accuracy and the comprehensibility of the generated rules, making learning from large data sets more difficult. Pruning is a...


Read the Article


Sorry, but this item is not currently available from your library.

Try another library?
Sign out of this library

Other Topics