Computational Intelligence Concepts To Implementations 1st Edition Russell C. Eberhart, Yuhui Shi - Solutions

Unlock the full potential of "Computational Intelligence Concepts To Implementations 1st Edition" by Russell C. Eberhart and Yuhui Shi with our comprehensive resources. Explore online answers key and solutions manual to enhance your understanding of complex concepts. Access solved problems and step-by-step answers with our chapter solutions and instructor manual. Our test bank offers a plethora of questions and answers for robust preparation. Enjoy the convenience of a solutions PDF and textbook, available for free download. Elevate your learning experience with expertly curated content designed for academic success.

What is U-Net, and what is it best used for?
Use a table to compare the characteristic features of AlexNet, Inception, VGG, ResNet, DenseNet, MobileNet, and EffecientNet.
Compare the features of three commonly used activation functions:REctified Linear Unit (ReLU), hyperbolic tangent, and the sigmoid function.
Explain the terms of convolutional layer, pooling layer, activation layer, dropout layer, and fully connected layer, in the context of convolutional neural networks.
What is a convolutional neural network?
What is an artificial neural network?
What is the difference between machine learning and deep learning?
What are the AutoML, PyCaret, and LazyPredict libraries?
What is ensemble learning?
What is Q-learning?
What is reinforcement learning, and what can it be used for?
What is semi-supervised learning?
What is K-means clustering?
What is unsupervised learning?
How does K-nearest neighbors work?
What is the difference between decision tree and random forest?
What is linear discriminant analysis, and what is principal component analysis?
What is naive Bayes, and how does it work?
What is SVM? What can it be used for?
What is the difference between classification and regression in supervised learning?
What is supervised learning?
What are popular Python AI frameworks?
What are commonly used AI datasets?
What is the standard input and output in a Python program?
What is Caffe2?
What is PyTorch?
What is Keras?
What is TensorFlow?
What is Scikit-Learn?
What is Scipy?
What are the most commonly used datasets in AI? Search the Internet and find five other datasets.
What is a Python virtual environment?
How is Python different from other programming languages such as C/C++ or Java?
What programming languages can be used for AI development? List the advantages and disadvantages of each language.
What types of computer hardware are available for AI development?Do an Internet search and find your ideal laptop/desktop computer for AI development.
Are you optimistic or pessimistic about the future of AI, and why?
Search the Internet and list your own “current state of AI” and give your own predictions for AI.
Search the Internet and list your own top 10 key moments of AI in history.
What are the differences between edge AI and cloud AI?
What are the three types of AI? What is the singularity?
What are AI hypes? Use a diagram to explain the different stages of an AI hype. What are AI winters?
Explain the differences of neural networks, machine learning, and deep learning.
Explain, with a suitable diagram, the three stages of AI development in history.
What is the Turing test, and what is the imitation game?
What is AI? Show a few examples of AI applications in your daily life.
8. Specify several possible codebook vectors for fuzzy expert systems
7. Using the Zurada sensitivity process, calculate the mean square average estimated sensitivity Si,eav for each input of the Iris dataset. What do these sensitivities tell you?
6. Propose an alternative way to display the information contained in a Hinton diagram.
5. Design and run experiments using PSO to plot relationships between the PSO's parameter (or input) values and its performance (or output) values.
4. Assume that you are implementing in the C language the method discussed in this chapter for calculating the Zurada sensitivity two for an m-input, n-output system. Draw the flowchart for your implementation.
3. Specify a method for calculating relation factor one for a fuzzy rule-based system.
2. Run the example software to explore the decision hypersurface between each pair of classes. Discuss your results.
1. Run the example software to obtain codebook vectors for each class of iris flower in the Iris dataset. Discuss your results.
7. Use PSO to evolve the weights for different configurations of neural networks that classify the Iris dataset. Use the Mann-Whitney U test to evaluate the configurations.Find two configurations for which the null hypothesis is not rejected and two for which it is rejected at the 0.05 level. Limit
6. Use k-fold cross validation to analyze the Iris dataset. Using 3-fold and 10-fold approaches, partition the dataset into 3 and 10 subsets, respectively. What results for training and testing are obtained with each approach? What significant differences do you see? For this dataset, what maximum
5. We are developing a simulation of the game of baseball. We want to measure how well our system simulates the margin of victory. In general, we like the margin of victory to be 1, 2, or 3 runs. A 4-run game isn't too bad, but games with margins of 5 or more runs are okay once in a while but
4. Given the following set of targets and outputs, calculate the average sum-squared error and the normalized error.Repeat your calculations, assuming that thetargets were 0.9 and 0.1 instead of 1 and 0, respectively, and the outputs were the same. Target Output 1 .90 I .87 1 .79 1 .89 1 .88 0 .12
3. Train a neural network on the Iris dataset using the back-propagation implementation(see Chapter 6). Plot an ROC curve for each of the three output PEs, using at least 10 values for the threshold value. Discuss your results.
2. For a pattern set we are using to train a neural network, one-half of the target values are 1 and one-half are 0. Calculate Emean. What happens to Emeanif targets of 0.9 and 0.1 are used instead of 1 and 0, respectively?
1. We have developed a diagnostic system with three possible diagnoses, Q, R, and S, that are equally probable (0.333 each). Assume that the costs associated with the three diagnoses are x, 2x, and 5x, respectively, and that the raw cost matrix comprises the base costs on the diagonal and the sum
7. How generally applicable is the system diagram of Figure 9.3 to other applications such as analysis of large video data streams? Identify another application area, and draw a diagram analogous to Figure 9.3 for that application.
6. Modify the fuzzy rule system in Listing 9.2, and run the software again to see whether you can obtain better results.
5. Run both the evolutionary fuzzy rule system (m_flaq = 0) and the fuzzy evolutionary fuzzy rule system (m_flaq - 1), and compare the results.
4. If you are asked to use a fuzzy system to adapt the PSO in the implementation of the evolutionary neural network discussed in Chapter 6, what will be the input and output of the fuzzy system?
3. Briefly describe how to use a fuzzy system to adapt the parameters you listed in Exercise 1.
2. Compare the strengths and weaknesses of the four levels of adaptation of genetic algorithms: environment, population, individual, and component.
1. List two parameters that can be adapted to improve a GA's performance at the levels of environment, population, individual, and component.
9. Repeat exercises 5 and 6, but use only 100 patterns to develop the fuzzy rules;then test (classify) all 150 patterns. Describe the differences in the results.
8. Repeat exercise 5, evolving everything: the rule set, the start and end points of each membership function, and the membership function types.
7. Repeat exercise 5, evolving both the rule set and the start and end points of each membership function.
6. Define five membership functions for each input variable and repeat exercise 3.Describe the differences in the results.
5. Run the evolutionary fuzzy rule system software on the Iris data using three membership functions for each input variable. Use all 150 patterns to develop the system. Evolve only the rule system. Set the acceptable performance to 0.949, which allows six errors out of the 150-pattern Iris dataset.
4. Redefine the FuzzyMember class so that it can be used to represent more general forms of membership functions such as an asymmetrical triangular function.
3. Define an enumeration data type for the methods to calculate the activation strength of the i f part for each rule, and specify the corresponding changes.
2. Define an enumeration data type for the membership functions used in the implementation, and illustrate what other changes should be made accordingly.
1. For a two-input, two-output fuzzy system, assume each variable has three fuzzy sets and encode the following rule in a string of bits: If Input_l is Medium and Input_2 is High, then Output_l is High and Output_2 is Low.
9. Following the five steps discussed in this chapter, design a fuzzy room temperature controller.
8. What are the advantages of using fuzzy controllers?
7. Implement the center of maximum method of defuzzification in the fuzzy source code.
6. Use the centroid with overlap defuzzification method to calculate the output value of the output variable FlowChange as shown in Figure 7.3.
5. Determine the percent of throttle applied for the conditions described in exercise 4.
4. Using the fuzzy membership functions defined in the chapter for the slowing of a train near a station, determine the percent of braking applied when the train is moving 3 km/hr and is 8 m from the station.
3. For each of the three measures of fuzziness defined in the chapter (equations 7.2, 7.3, and 7.4), calculate the fuzzy entropy of one of the fuzzy membership functions you defined in exercise 1. (Remember that the membership function is defined over the entire age domain.)
2. Fuzzy sets V and W are defined on the same universe of five individuals as follows:For Vand W, find: (a) VA W, (b)Vu W, (c)V, (d)W, (e)Vn W, (f)Vu W. V = 1.0 0.8 0.6 0.20 {+! w= {} W r + 1.0 0.6 + + T $ t 0.20 + 0 }) 0.45 0.15 + + S t
1. Given that we are working with the domain of "age" of a population of people, define a set of fuzzy membership functions over the domain that might be appropriate for use by an insurance company determining risk of Alzheimer's disease, which affects mainly older people. Repeat the exercise for
17. Run the learning vector quantization implementation on the web site using the run files provided. Vary the number of clusters specified. Try 6 clusters as specified in the run file on the Web; then try 3 clusters, then try 9 clusters. Discuss your results. Which number of clusters would you
16. Run the back-propagation implementation on the web site using the run files provided for training and testing on the Iris dataset. Keeping everything else constant in the run files, try training for 10,000, 1,000, and 100 iterations. Discuss your results with respect to both training and
15. Modify the evolutionary back-propagation implementation so that it also evolves the slope of the sigmoid function.
14. Based on the implementation from exercise 8, implement the SOFM with a two-dimensional output slab.
13. Modify the LVQ source code for the implementation of a SOFM with a one- dimensional output slab.
12. Define one or two decreasing functions of time for updating the learning rate (t) and make corresponding changes in the 1vq_update_learning_rate() routine. Compile and run the LVQ and compare the performance with the original version.
11. In the state BP FEEDFORWARD_OUTPUT, modify the state handling routine so that calculations for only one output PE are done for each cycle of the state machine.
10. In the state BP_FEEDFORWARD_HIDDEN, all hidden layers are calculated in one cycle of the state machine. Modify the code so that only one hidden layer is calculated in one cycle.
9. Assume that the activation function for PEs in the hidden and output layers of the back-propagation implementation is the hyperbolic tangent function. Specify the changes that should be made in the source code.
8. Using BP_Neuron_Type as the definition for the PE (neuron) in the back-propagation network, make necessary changes to other data types, and specify the corresponding changes that should be made in the source code.
7. Add an error termination criterion into the back-propagation implementation so that the BP training can be terminated based on either maximum number of generations or the error termination criterion.
6. List the attributes necessary to specify a learning vector quantizer neural network implementation. Repeat for a self-organizing feature map implementation.
5. List the attributes necessary to specify a back-propagation neural network implementation. Repeat for an evolutionary back-propagation implementation.
4. Compare the LVQ-I, LVQ-II, and SOFM network paradigms. What are the similarities? What are the differences? Why might you choose to use SOFM rather than LVQ-I?
3. It is usually recommended to scale or preprocess the input values prior to presentation to a neural network. List some reasons for doing this. List some ways to do this.

Showing 3100 - 3200 of 4756

First
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
Last

Computational Intelligence Concepts To Implementations 1st Edition Russell C. Eberhart, Yuhui Shi - Solutions

Step by Step Answers