S. M. Weiss and C. A. Kulikowski, Computer Systems That Learn. San Mateo, CA: Morgan Kauffman, 1990. Home/Search Document Not in Database Summary Related Articles Check |
This paper is cited in the following contexts:
First 50 documents Next 50
GRASP: Recognition of Australian Sign Language Using Instrumented .. - Kadous (1995) (3 citations) (Correct)
.... of decision trees is approximately logarithmic in the number of training instances (because of the tree structure employed for classification) and similarly, classification time for instance based learning techniques can be made logarithmic in the number of training instances, using k d trees ([WK91]) Learning times for simple instance based learning techniques is linear, and it has been shown that under certain circumstances, so is C4.5 ( Squ94] If we have more samples, we can simply add them in, without the major penalty of a quadratic increase in the time cost of learning. Furthermore, ....
....can be done very quickly. It is unusual for a tree to have a depth greater than 20, so there is a nice upper bound. IBL, however, must find the instance that is closest to it. If a brute force algorithm is used, then it has to check every instance in the data set. Fortunately, as illustrated in [WK91], algorithms exist that allow this search in O(log n) manner, using a method known as k d trees. C4.5 is only one example of a symbolic decision tree learner. The implementation of the IBL algorithm used does not employ k d tree. It uses an optimised search based on collecting all instances of ....
Sholom M. Weiss and Casimir A. Kulikowski. Computer Systems That Learn. Morgan Kaufman, 1991.
Detection of Anti-Personnel Land-Mines Using.. - Cremer.. (1999) (Correct)
....of representing the training set. For instance, the rule based method has more free parameters than other methods, so it can possibly better represent the training set. For an unbiased comparison of these methods, with the limited data set we have, a leave oneout evaluation method is used, see [18]. In the leave one out evaluation method, the parameters for each method are acquired on a training set, which contains all but one sample (a region containing one mine and on average 1 ) The acquired parameters from the training set are tested on the single sample left out (the evaluation ....
S. Weiss and C. Kulikowski, Computer systems that learn. Morgan Kaufmann Publishers, 1991.
Inductive Specification Recovery: - Understanding Software By (Correct)
....databases. If we report a recall of 50 for v this means that on average half of the specifications that are consistent with all of the sample databases can be obtained by running the learner on a single database. This procedure is broadly similar to using cross validation to measure accuracy [Weiss and Kulkowski, 1990]; however, rather than measuring the predictive accuracy of a hypothesis on hold out data, a hypothesis is evaluated by seeing if it is 100 correct on a set of holdout databases. This rewards the hypotheses most valuable in a discovery context: namely, the hypotheses that are with high ....
Sholom Weiss and Casmir Kulkowski. Computer Systems that Learn. Morgan Kaufmann, 1990. 24
Processing of Polarimetric Infrared Images for Landmine.. - Cremer, de Jong, Schutte (2003) (Correct)
....of influence. Each detection is of limited size. To compare results presented here, each false alarm can be seen as one SCOOP false alarm. The data set is limited with 60 landmines and around 20 m . Although, this data set is larger than in previous sensor fusion experiments [1] leave one out [16] has been used to evaluate the performance on an independent data set. The advantage of leave one out is that as much training data is used as possible. The disadvantage of leave one out is that optimisation of a complete ROC curve is not straight forward. One approach is to remove landmines in a ....
S. M. Weiss and C. A. Kulikowski, Computer systems that learn. Morgan Kaufmann Publishers, 1991.
Grammatically Biased Learning: Learning Horn Theories Using an.. - Cohen (1991) (2 citations) (Correct)
....predicates that were used in [ Quinlan, 1990b ] to solve the same problems using the FOIL system. As it turns out, similar typing, symmetries, and predicate combinability constraints can also be used in this domain. Since the amount of data is small, we used the leave one out technique [ Weiss and Kulkowski, 1990 ] to estimate error rates. For a layout with n cards, n runs of each learning system were made, where in each run, one of the n examples was withheld during training and used as a test case. The average error rates for these n runs was used as an estimate of the true error rate of the learning ....
....in Prolog. However, some of the difference is probably due to the overhead of using tuple based measures of hypothesis quality rather than the simpler example based measures of quality employed by C4.5. We also note that many other learning techniques show good performance on this dataset [ Weiss and Kulkowski, 1990, pages 152 153 ] As another aside, the experiments with the iris data are also interesting because they are a real world dataset, whereas most of the experiments described in this paper deal with artificial learning problems. To summarize the results of this section, although we have not ....
Sholom Weiss and Casmir Kulkowski. Computer Systems that Learn. Morgan Kaufmann, 1990.
MERBIS - A Multi-Objective Evolutionary Rule Base Induction.. - Setzkorn, Paton (2003) (Correct)
....of a classifier has practical importance. Some researchers argue that only comprehensible classifiers are actually adopted in practice [27, 28, 41] One reason for this might be that domain experts are very wary and distrustful of the incomprehensible results generated by a computer [53]. Comprehensibly can be achieved by symbolic classifiers, which correspond to an explicit knowledge representation form [1, 39] MERBIS produces Fuzzy Classification Rule Bases (FCRBs) which are a specific type of symbolic classifiers, which can exhibit high transparency [35] The performance of ....
S. Weiss and C. Kulikowski. Computer Systems That Learn. Morgan Kaufmann, 1991.
A Comparison of Decision-Level Sensor-Fusion.. - Cremer, Schutte.. (2001) (1 citation) (Correct)
....optimal solutions for other data sets and other methods is larger, so only one (arbitrary) of the solutions will be evaluated in the experiments. 3.4. Leave one out For an unbiased comparison of the fusion methods, with the limited data set we have, a leave one out evaluation method is used [44]. In the leave one out evaluation method, the parameters for each method are acquired on a training set, which contains all but one sample (a region containing one landmine) The acquired parameters from the training set are tested on the single sample left out (the evaluation set) This is ....
S.M. Weiss, C.A. Kulikowski, Computer Systems that Learn, Morgan Kaufmann, Los Altos, CA, 1991.
Direct Marketing Performance Modeling Using Genetic Algorithms - Bhattacharyya (1999) (1 citation) (Correct)
....two, potentially conflicting, objectives. Overfitting to the training data and consequent shrinkage in performance from the training to the test data is common in application of machine learning techniques. In an attempt to further control such shrinkage, we investigate the use of resampling [2, 8, 30] within the genetic learning process and a second set of experiments examines performance enhancements through resampling based learning. The following section first looks at performance analysis of DM models and presents the proposed modeling approach. Section 2 elaborates on the use of genetic ....
....to control shrinkage from the training to the test data, a second set of experiments undertakes a preliminary investigation of resampling incorporated into the GA learning process. Resampling techniques, utilized mainly as a bias reduction tool in the estimation of error rates in classifiers, [2, 30] offer potential advantages to GA based learning of robust models. Here, the fitness of population members is estimated as an average of fitness values obtained from multiple subsamples of the training data. This biases the search towards solutions that perform uniformly well across the different ....
S.M. WEISS and C.A. KULIKOWSKI, 1991. Computer Systems that Learn, Morgan Kaufmann, San Mateo, CA.
Computational Intelligence Methods for Rule-Based Data.. - Duch, Setiono, Zurada (2004) (Correct)
No context found.
S. M. Weiss and C. A. Kulikowski, Computer Systems That Learn. San Mateo, CA: Morgan Kauffman, 1990.
Knowledge Acquisition by Symbolic Tree Induction for.. - Perner, al. (1996) (Correct)
No context found.
S. M. Weiss and C.A. Kulikowski, " Computer Systems that learn," Morgan Kaufmann, 1991.
Unknown - (Correct)
No context found.
S. M. Weiss and C. A. Kulikowski, Computer systems that learn, Morgan Kaufmann, San Mateo, CA, (1991).
Neural Network Recognition of - Hand-Printed Characters Sameer (Correct)
No context found.
S. M. Weiss and C. A. Kulikowski, Computer Systems that Learn, Kauffman, CA, 1991.
Visual Object Categorization Using Distance-based.. - Kosinov.. (2004) (Correct)
No context found.
S. Weiss and C. Kulikowski. Computer Systems That Learn. Morgan Kaufmann, 1991.
Effects of Three-Objective Genetic Rule Selection on the.. - Ishibuchi, Yamamoto (2003) (Correct)
No context found.
Weiss, S. M., and Kulikowski, C. A.: Computer Systems That Learn, Morgan Kaufmann Publishers, San Mateo (1991).
Feature level fusion of polarimetric infrared and.. - Cremer, de Jong.. (2003) (Correct)
No context found.
S. M. Weiss and C. A. Kulikowski, Computer systems that learn. Morgan Kaufmann Publishers, 1991.
Applying Genetic and Symbolic Learning Algorithms to .. - Milaré, Batista.. (2004) (Correct)
No context found.
S. M. Weiss and C. A. Kulikowski. Computer Systems that Learn. Morgan Kaufmann, 1991.
Graphical Methods for Classifier Performance Evaluation - Batista (2003) (Correct)
No context found.
S. M. Weiss and C. A. Kulikowski. Computer Systems that Learn. Morgan Kaufmann, San Mateo, CA, 1991.
Computational Intelligence Methods for Rule-Based Data.. - Duch, Setiono, Zurada (2004) (Correct)
No context found.
S. M. Weiss and C. A. Kulikowski, Computer Systems That Learn. San Mateo, CA: Morgan Kauffman, 1990.
MERBIS - A Self-Adaptive Multi-Objective - Evolutionary Rule Base (2003) (Correct)
No context found.
S. Weiss and C. Kulikowski. Computer Systems That Learn. Morgan Kaufmann, 1991.
Effect of Rule Weights in Fuzzy Rule-Based Classification.. - Ishibuchi, Nakashima (2000) (1 citation) (Correct)
No context found.
S. M. Weiss and C. A. Kulikowski, Computer Systems That Learn, Morgan Kaufmann, San Mateo, 1991.
Linear and Order Statistics Combiners for Pattern Classification - Tumer, Ghosh (1999) (21 citations) (Correct)
No context found.
S. M. Weiss and C.A. Kulikowski. Computer Systems That Learn. Morgan Kaufmann, 1991.
S E A R C H P O R T I D I A P D a l l e M o l l e I n s t i t u t .. - Pe Cep Ua (2002) (Correct)
No context found.
S.M. Weiss and C.A. Kulikowski. Computer Systems That Learn. Morgan Kaufmann, 1991.
An empirical study of the use of relevance information in.. - Srinivasan, al. (2003) (Correct)
No context found.
S.M. Weiss and C.A. Kulikowski. Computer systems that learn. Morgan Kaufmann, San Mateo, CA, 1991.
Comparison of vehicle-mounted forward-looking.. - Cremer.. (2003) (Correct)
No context found.
S. M. Weiss and C. A. Kulikowski, Computer systems that learn, Morgan Kaufmann Publishers, 1991.
Towards an Operational Sensor-Fusion System for.. - Cremer, Schutte.. (2000) (Correct)
No context found.
S. M. Weiss and C. A. Kulikowski, Computer systems that learn, Morgan Kaufmann Publishers, 1991.
First 50 documents Next 50
Online articles have much greater impact More about CiteSeer.IST Add search form to your site Submit documents Feedback
CiteSeer.IST - Copyright Penn State and NEC