Training and Test Data

The selection of training data presented to the neural network influences whether or not the network learns a particular task. Some major considerations include the generalization/memorization issue, the partitioning of the training and prediction sets, the quality of data, the ratio of positive and negative examples, and the order of example presentation. 

---

Divide the available data into training and test data sets (1/3). Test sets are used to validate the trained network and insure accurate generalization. 

Cross validation and bootstrap techniques can be applied for a statistically based estimation of the optimum number of PCA components. The idea is to randomly split the data into training and test data. PCA is then applied to the training data and the observations from the test data are reconstmcted using 1 to m PCs. The prediction error to the real test data can be computed. Repeating this procedure many times indicates the distribution of the prediction errors when using 1 to m components, which then allows deciding on the optimal number of components. For more details see Section 3.7.1. 

The above example demonstrates that the choice of k is crucial. As mentioned above, k should be selected such that the smallest misclassification rate for the test data occurs. If no test data are available, an appropriate resampling procedure (CV or bootstrap) has to be used. Figure 5.14 shows for the example with three overlapping groups, used above in Figure 5.12, how k can be selected. Since we have independent training and test data available, and since their group membership is known, we... 

FIGURE 5.14 k-NN classification for the training and test data used in Figure 5.12. The left plot shows the misclassification rate for the test data with varying value of k for k-NN classification, and the right plot presents the result for k = 25. The misclassified objects are shown by dark symbols. 

The number of neurons in the hidden layer was therefore increased systematically. It was found that a network of one hidden layer consisting of twenty neurons, as shown in Figure 2.6, performed well for both the training and testing data set. More details about the performance of this network will be given later. The network architecture depicted in Figure 2.6 consists of an input layer, a hidden layer, and an output layer. Each neuron in the input layer corresponds to a particular feed property. The neurons... 

Examined below are several neural network design considerations, including the architecture (8.1), learning algorithm (8.2), network parameters (8.3), training and test data (8.4), and evaluation mechanism (8.5). 

Prediction Accuracy of the Transmembrane Segments on the Training and Testing Data Sets... 

To test the classification performance of the adaptive wavelet, the coefficients from each of the bands (at level 2) at initialization and at termination of the algorithm were used as inputs to the classifier. The results are summarized for both the training and test data in Table I. At initialization the coefficients in band(2,0) gave the best classification rates closely followed by band(2,l). At completion the classification performance of band(2,0) has further improved. 

One might be interested in how the adaptive wavelet performs against predefined filter coefficients. In this section, we perform the 2-band DWT on each data set using filter coefficients from the Daubechies family with Nf = 16. The coefficients from some band (j, x) are supplied to BLDA. We consider four bands - band(3,0), band(3,l), band(4,0) and band(4,l). The results for the training and testing data are displayed in Table 3. The test CCR rates are the same for the seagrass and butanol data, but the AWA clearly produces superior results for the paraxylene data. 

Fig. 3.60 The Eureqa interface with the caco-2 training and test data loaded...

Take, for example, the training and test data shown in Fig. 10.12. Overtraining results in the fit shown with a dashed line, where the points in the test set are poorly fitted to the line. The ideal line is shown by a solid line. 

Figure 1. Land slide s triggered by the Yushu earthquake, training and testing data for ANN modeling.

The proposed model is a feed-forward GMDH-type network and has constructed using experimental data set from ref. (Goncalves et al., 2002). This data set is constituted of 25 points in four different concentrations of water in solvent In Table 6.1, the overall experimental compositions of the mixtures and in Table 6.2, experimental mass fractions of the components in alcohol and oil phase are shown. The data set is divided in two parts, 80% used as training and 20% used as testing data. Each point in training and test data is constituted of 13 values. The 4 mass fractions in overall compositions and water concentration in solvent are normalized and used as inputs of GMDH-type network (X, and other 8 values are used as desired outputs of network, 4 mass fractions in alcohol phase. .., Y ) and 4 mass fractions in oil phase (Z, ... 

---