Setiono, R., and Liu, H. Neural network feature selectors. IEEE Trans. On Neural Networks, 8(3), 654-662 (1997).  Home/Search   Document Details and Download   Summary   Related Articles   Check

....of groups, called factors, that reflect the major dimensions of the phenomenon under consideration. A genetic algorithm is used to explore the feature space originated by the factors and to determine the set of the most informative feature configurations. Neural Network Feature Selector (NNFS) [9] is a method that adds a penalty term to the error function used to derive the weight updating rule of neural network training. GADistAl [10] is a wrapper based approach to feature selection using a genetic algorithm in conjunction with a constructive neural network learning algorithm called ....

....the number of features and then runs Automatic Branch Bound (ABB) a complete search algorithm. B. Comparisons The algorithms that are surveyed for the comparison including the original wrapper [1] Las Vegas filter(LVF) 13] Las Vegas wrapper(LVW) 4] neural net feature selector (NNFS) [9], hybrid approach(hybrid) 12] informationtheoretic filter(INFO) 14] and AHOC genetic algorithm (AHOC) 8] The last column of Table III shows these performance data. The comparison is made on the basis of both the number of features selected and the error rates after feature selection. It ....

R. Setiono and H. Liu, "Neural network feature selector," IEEE Transactions on Neural Networks, vol. 8, no. 3, 1997.

....variations of these algorithms [10] The algorithms generally work very well in many cases. However, they have the common characteristics that they ignore the fact that some features may be relevant only in a context. Setiono Liu have proposed an neural network based feature selection method [6] Their method is based on a three layer feed forward neural network, which is trained with conventional learning algorithms, but they uses an augmented error function in the learning process. The method uses the network to select the input attributes that are most useful for discriminating ....

....from the Merikoski Institute of Health Research and Rehabilitation, Oulu, Finland. The material included 305 R R interval measurements and accurate oxygen uptake measurements (with an ergospirometer) of the subjects who participated in the study. The subjects were adult men and women aged 15 65 years. All the subjects were healthy and none of them had any medication. The fitness level was indicated as a fitness class of 1 to 5. The distribution of the research material is shown in Fig. 2. The reason why, good fitness levels (class 5) were so common, was that quite a significant portion ....

[Article contains additional citation context not shown here]

Setiono, R., Liu, H., "Neural-Network Feature Selector", IEEE Transactions on Neural Networks, 8, (3), 654-662. 1997.

....function of all the outputs in the network. Ruck et al. [12] proposed a node saliency measure that analysed the sensitivity of the network outputs with changing inputs. If the outputs changed dramatically for a small input change this feature was considered important to the problem. Setiono et al. [14] suggested that instead of using a saliency measure which is a function of the network weights that one could use the network classification performance on a validation dataset directly. In his algorithm the drop in the error on a verification dataset was observed whilst setting different weight ....

R Setiono and H Liu. Neural network feature selector. IEEE Transactions on Neural Networks, 8(3):654--661, May 1997.

....all datasets, while BSE is effective for datasets with a large number of original attributes. ffl The ANNIGMA wrapper approach is computationally feasible. Recently, many clever feature selection techniques for neural nets were proposed, ranging from filter based approaches to genetic algorithms [8, 9, 10, 11, 12, 13]. Comparing our experimental results with the performance data reported in their papers, we found surprisingly that our simple approach outperforms their sophisticated approaches in almost all test datasets. This suggests that feature selection for neural nets might not be as difficult as ....

.... the UCI Machine Learning Repository [7] These datasets were chosen to include datasets with various characteristics and to maximize the comparability of the ANNIGMA wrapper approach with other published approaches to feature subset selection, especially those concerning neural network induction [6, 8, 10, 11] 14 3.1 Dataset Description and Preparation Table 1 gives the summary of our experimental datasets. The second column lists the amount of samples for training and the third column lists the data sizes of the hold out sets for validation. The ratio of training and hold out set is two to one ....

[Article contains additional citation context not shown here]

R. Setiono and H. Liu, "Neural network feature selector," IEEE Transactions on Neural Networks, vol. 8, no. 3, pp. 654--662, 1997.

....of the GA requires training a neural net for each fitness evaluation performed by the GA. Due to the stochastic element of neural net training, each net should be trained several times with different initial weights in order to properly assess performance; at least 30 repeats has been suggested in [14]. This makes this approach computationally prohibitively expensive for all but trivial problems. The solution has usually been to use a simple approximation as the GA objective function [3] or to only partially train a subset of the neural nets in each generation [6, 17] A different and more ....

R. Setiono and H. Liu. Neural-network feature selector. IEEE Transactions on Neural Networks, 8(3):654--659, 1997.

....the relevance of a single variable is thus (almost) equal to the time needed to process a dataset by the neural network. The data modification algorithms can be separated into three different groups. Constant substitution substitutes a constant value for the input variable under investigation [2, 3, 5, 6, 7, 8], translation factor modifies the data by translation [7, 9, 10] and data permutation permutes the data of input variable i across patterns [11] 3.2 Missing values The following algorithms treat the removed input variable as a missing value and approximate the performance without an input ....

Rudy Setiono and Huan Liu. Neural-network feature selector. IEEE Transactions on Neural Networks, 8(3):654--662, May 1997.

....of the GA requires training a neural net for each fitness evaluation performed by the GA. Due to the stochastic element of neural net training, each net should be trained several times with different initial weights in order to properly assess performance; at least 30 repeats has been suggested in [7]. This makes this approach computationally prohibitively expensive. The solution has usually been to use a simpler, related classifier in the GA [1] or to only partially train a subset of the neural nets in each generation [3, 11] A different approach, described below, is to combine the ....

Setiono, R. & Liu, H.(1997). Neural-network feature selector. IEEE Transactions on Neural Networks 8(3): 654--659.

No context found.

R. Setiono and H. Liu, "Neural network feature selector," IEEE Trans. Neural Networks, vol. 8, pp. 654--662, May 1997.

....of useful input units corresponds to the number of relevant input attributes of the data. Typical algorithms usually start by assigning one input unit to each attribute, train the network with all input attributes and then remove network input units that correspond to irrelevant data attributes [9, 10]. Various measures of the contribution of an input attribute to the network s predictive accuracy have been proposed [11, 12, 13, 14, 15] We have opted for the destructive approach since in addition to producing a network with the fewest hidden units, we also wish to remove as many redundant and ....

R. Setiono and H. Liu, \Neural network feature selector," IEEE Trans. on Neural Networks, vol. 8, no. 3, pp. 654-662, 1997.

....of useful input units correspond to the number of relevant input attributes of the data. Typical algorithms usually start by assigning one input unit to each attribute, train the network with all input attributes and then remove network input units that correspond to irrelevant data attributes [15, 16]. Various measures of the contribution of an input attribute to the network predictive accuracy have been developed [2, 10, 13, 18] The purpose of this paper is (1) to present an algorithm for removing redundant or irrelevant input and hidden units from feedforward neural networks for regression ....

Setiono, R. and Liu, H. (1997) Neural network feature selector. IEEE Trans. on Neural Networks, 8 (3), 654-662.

....it is certain that it will fail on problems whose attributes are highly correlated such as the parity problem where the combinations of a few attributes do not help in finding the relevant attributes. One solution to this problem is to use feedforward neural networks as a feature selector (Setiono Liu, 1997). The idea is to train a network and then prune it while maintaining its performance. At the end, features with connections to hidden units are chosen as selected features. Another common understanding is that some learning algorithms have built in feature selection, for example, ID3 (Quinlan, ....

Setiono, R., & Liu, H. (1997). Neural network feature selectors. IEEE Trans. on Neural Networks, 8 (3), 654-662.

No context found.

Setiono, R., and Liu, H. Neural network feature selectors. IEEE Trans. On Neural Networks, 8(3), 654-662 (1997).

No context found.

Rudy Setiono. Neural network feature selector. IEEE Trans. Neural Networks, 8(3):654--662, 1997. 44

No context found.

Setiono, R. & Liu, H. 1997. 'Neural-network feature selector', IEEE Transactions on Neural Networks vol. 8, no. 3, p.654 -- 662.

No context found.

R. Setiono and H. Liu, \Neural-network feature selector," IEEE Trans. of Neural Networks, vol. 8, no. 3, pp. 654-662, 1997.

No context found.

, 654-662.

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