### Table 2 : Neural Network Configurations

1997

"... In PAGE 3: ...001. Very simple ANN configurations were chosen (see Table2 ) using as few neurons as possible. To train the network to the desired precision, several thousand iterations were required in each case taking just few seconds to complete.... ..."

Cited by 1

### Table 1: Joint Probability Distribution

"... In PAGE 4: ... Probabilistic/Statistical Reasoning: Many forms of information are inherently probabilistic | eg, given certain symptoms, we may be 80% con dent the patient has hepatitis, or given some evidence, we may be 10% sure a speci c stock will go up in price. One possible downside of dealing with probabilities is the amount of information that has to be encoded: in general one may have to express the entire joint distribution, which is exponential in the number of features; see Table1 . For many years, this observation motivated researchers to seek ways to avoid dealing with probabilities.... In PAGE 4: ... To make this concrete, consider the claims that Hepatitis \causes quot; Jaundice and also \causes quot; a Bloodtest to be positive, in that the chance of these symptoms will increase if the patient has hepatitis. We can represent this information using the full joint over these three binary variables (see Table1 for realistic, if fabricated, numbers), then use this information to compute, for example, P( H j :B ) | the posterior probability that a patient has hepatitis, given that he has a negative blood test. The associated computation, P( H j :B ) = P( H; :B ) P( :B ) = PX62fH;Bg Px2X P( H; :B; X = x ) PX62fBg Px2X P( :B; X = x ) involves the standard steps of marginalization (the summations shown above) to deal with unspec- i ed values of various symptoms, and conditionalization (the division) to compute the conditional probability; see [Fel66].... In PAGE 5: ... While the saving here is relatively small (2 links rather than 3, and a total of 5 parameters, rather than 7), the savings can be very signi cant for larger networks. As a real-world example, the complete joint distribution for the Alarm belief net [BSCC89], which has 37 nodes and 47 arcs, would require approximately 1017 parameters in the naive tabular representation | a la Table1 . The actual belief net, however, only includes 752 parameters.... In PAGE 6: ... There are many obvious connections between the logic-based and probability-based formalisms. For example, we can view the possible worlds (as shown in Table 2) as a \qualitative quot; version of the atomic events (see Table1 ), with the understanding that each \impossible quot; world has probability 0 of occuring, and the other possible words have non-0 probability. Many, including Nilsson [Nil86], have provided formalisms that attempt link these areas.... ..."

Cited by 3

### Table 14-7: Neural network precision (standard deviation) and recall.

"... In PAGE 3: ... DATA ACQUISITION The development of haptic devices is in its infancy. We have focused our research and experiments on the CyberGrasp exoskeletal interface and accompanying CyberGlove, which consists of 33 sensors ( Table14 -1). We use the CyberGrasp SDK to write handlers to record sensor data for our experiments whenever a sampling interrupt occurs.... In PAGE 3: ... We term each of these 10 letters a sign. The 22 sensor values (excluding sensors 23 to 33 in Table14 -1) are recorded in a log file for each sign made by a subject, termed as a session. Each session log file con- tains thousands of rows of sensor values sampled at some frequency, which depends on the sampling technique used.... In PAGE 4: ...241 Table14 -1: CyberGrasp sensors. Sensor Num- ber Sensor Description Sensor Num- ber Sensor Description 1 Thumb roll sensor 15 Ring middle abduction 2 Thumb inner joint 16, 17, 18 Pinky inner, middle, outer joint 3 Thumb outer joint 19 Pinky ring abduction 4 Thumb index abduction 20 Palm arch 5, 6, 7 Index inner, middle, outer joint 21 Wrist flexion 8, 9, 10 Middle inner, middle, outer joint 22 Wrist abduction 11 Middle index abduction 23, 24, 25 x, y, z location 12, 13, 14 Ring inner, middle, outer joint 26, 27, 28 x, y, z abduction 29 to 33 Forces for each finger Sampling Techniques To record several snapshots for each static sign made within a session, we need to sample the values of sensors for each subject making a sign.... In PAGE 14: ...92). Table14 -2 illustrates a comparison among the techniques. Table 14-2: Overall classification error.... In PAGE 14: ....92). Table 14-2 illustrates a comparison among the techniques. Table14 -2: Overall classification error. Error Standard Derivation C4.... In PAGE 14: ...e., C, G, and H ) quite well (see Table14 -3). Note that although C4.... In PAGE 15: ... Understanding of User Behavior in Immersive Environments Chapter 14 Bayesian Classifier decides based on probability distribution of the input samples, it tends to perform quite well overall despite intuitive variations in performance of signs by different subjects. Table14 -3: Best recognition technique for each sign. A B C D E F G H I L C4.... In PAGE 15: ...N C4.5 C4.5, B C4.5 Table14 -4: Nearest neighbors for each sign in multidimensional space. Nearest Farthest Avg.... In PAGE 15: ...693604 L G D A E H C I F B 2.697530 As illustrated in Table14 -8,1 we see that the best classifier for a sign is not necessarily the one that confuses the sign with fewer other signs. The decision to choose a classifier from given classifiers becomes very much application-dependent.... In PAGE 16: ...253 Bayesian Classifier and an appropriate I/O design we can achieve an acceptable perform- ance. Table14 -5: C4.5 Precision and recall.... In PAGE 16: ... 0.00 0.00 0.00 6.66 6.66 0.00 6.66 0.00 0.00 80.00 Table14 -6: Bayesian precision and recall. A B C D E F G H I L A 86.... In PAGE 17: ... Table14 -8: Number of other signs with which each sign is confused for different classifiers. Sign Bayesian C4.... ..."

### Table 1 Analogies between neural networks and atlas-based classifiers Neural network

2004

"... In PAGE 2: .... Rohlfing, C.R. Maurer Jr. / Pattern mented using three individual atlases, or using a generic classifier, say a neural network, and an atlas-based classifier. Table1 gives an overview of the related concepts. From an operational per- spective, an atlas-based classifier takes as its input a coordinate x inside the domain of the unseg- mented target image, in the 3D case R3, and pro- duces as its output the label assigned to that coordinate.... ..."

### Table 7: Recommendations for Neural Network Use with Education Policy Analysis Questions

in Enhancing our Understanding of the Complexities of Education: "Knowledge Extraction from Data" using

"... In PAGE 22: ... (See Table 6) Table 6: Over and Under-representation of Asian/Pacific Island Students Group CHI FIL JAP KOR SEA PI SA WA ME OTH 1 -1% 3% -2% -4% 6% 4% -5% -1% -1% 1% 2 -1% 1% 4% -9% 5% 1% -1% -2% 4% -1% 3 0% 0% 1% 1% -3% -3% 3% 1% 1% -1% 4 9% -5% 1% 2% -7% 0% 0% -2% -2% 4% 5 -2% -3% -1% 6% -1% 0% 2% 0% -1% -1% Si milar discrepancies appear among Hispanic subgroups. Table7 suggests that the pattern of representation of the Hispanic aggregate group was substantially driven by the distribution of Mexican (MEX) students. Cuban students, to the contrary, were more likely to be found grouped with Asian/Pacific Island or White students than their Hispanic, Mexican counterparts.... In PAGE 22: ... Cuban students, to the contrary, were more likely to be found grouped with Asian/Pacific Island or White students than their Hispanic, Mexican counterparts. Table7 : Over and Under-representation of Hispanic Students Group MEX CUB PR OTHH 1 4.3% -1.... In PAGE 23: ... Yet, similar problems are likely to occur even when conventional methods are used. Table7 provides rough guidelines for applying neural networks to problems or questions related to education policy. Broadly speaking, the first two studies presented in this paper point to the particular value of hybrid neural/regression methods that apply neural or genetic algorithm estimation techniques to identify or construct a best predicting non-linear regression equation.... ..."

### Table 6: Neural network results.

"... In PAGE 5: ... Figure 5: Signal space for neural network. Table6 shows that the network using the 7- signal characteristic set gave the correct result 93.... ..."

### Table 4: Neural network tools.

2003

"... In PAGE 20: ... They are applicable in almost every situation where a relationship between input and output parameters exists, even in cases where this relationship is very complex and cannot be expressed or handled by mathematical or other modelling means. Table4 summarizes the features of the most commonly used neural network tools. Beyond general purpose and stand-alone tools, there exist library tools, such as the SPRLIB and the ANNLIB (developed by the Delft University Technology at Netherlands) emphasizing on image classification and pattern recognition applications.... ..."

Cited by 2

### Table 8 Neural network parameters

"... In PAGE 30: ... The successful ANN models were saved in a file along with information about that particular model, such as variable selection, variable transformations, and number of hidden nodes established. Table8 gives network parameters used to build the ANN models. Testing the ANN Models The ANN models were tested by running them using the test data sets prepared for each element.... ..."

### Table 1 Neural network architectures

2003

"... In PAGE 6: ...etter as the scale is increased, i.e. as the data becomes smoother. On the final smooth trend curve, resid(t)in Table1 , a crude linear extrapolation estimate, i.e.... In PAGE 6: ...avelet coefficients at higher frequency levels (i.e. lower scales) provided some benefit for estimating variation at less high frequency levels. Table1 sum- marizes what we did, and the results obtained. DRNN is the dynamic recurrent neural network model used.... In PAGE 6: ...sed. The architecture is shown in Fig. 3. The memory order of this network is equivalent to applying a time- lagged vector of the same size as the memory order. Hence the window in Table1 is the equivalent lagged vector length. In Table 1, NMSE is normalized mean squared error, DVS is direction variation symmetry (see above), and DS is directional symmetry, i.... In PAGE 6: ... Hence the window in Table 1 is the equivalent lagged vector length. In Table1 , NMSE is normalized mean squared error, DVS is direction variation symmetry (see above), and DS is directional symmetry, i.e.... In PAGE 7: ...ion of these results can be found in Ref. [4]. For further work involving the DRNN neural network resolution scale. From Table1 , we saw how these windows were of effective length 10, 15, 20, and 25 in terms of inputs to be considered. Fig.... ..."

### Table 7 provides rough guidelines for applying neural networks to problems or

in Enhancing our Understanding of the Complexities of Education: "Knowledge Extraction from Data" using

"... In PAGE 22: ... (See Table 6) Table 6: Over and Under-representation of Asian/Pacific Island Students Group CHI FIL JAP KOR SEA PI SA WA ME OTH 1 -1% 3% -2% -4% 6% 4% -5% -1% -1% 1% 2 -1% 1% 4% -9% 5% 1% -1% -2% 4% -1% 3 0% 0% 1% 1% -3% -3% 3% 1% 1% -1% 4 9% -5% 1% 2% -7% 0% 0% -2% -2% 4% 5 -2% -3% -1% 6% -1% 0% 2% 0% -1% -1% Si milar discrepancies appear among Hispanic subgroups. Table7 suggests that the pattern of representation of the Hispanic aggregate group was substantially driven by the distribution of Mexican (MEX) students. Cuban students, to the contrary, were more likely to be found grouped with Asian/Pacific Island or White students than their Hispanic, Mexican counterparts.... In PAGE 22: ... Cuban students, to the contrary, were more likely to be found grouped with Asian/Pacific Island or White students than their Hispanic, Mexican counterparts. Table7 : Over and Under-representation of Hispanic Students Group MEX CUB PR OTHH 1 4.3% -1.... In PAGE 25: ...Page 25 of 28 Table7 : Recommendations for Neural Network Use with Education Policy Analysis Questions Type of Problem Preferred NN(s) Notes Conventional Methods (Comparison/Validation Time Series Prediction (Multivariate) 1. GMDH 2.... ..."