4. Meta Model Describe how to use stacking, to create a meta model. 5. Classifie
ID: 653235 • Letter: 4
Question
4. Meta Model Describe how to use stacking, to create a meta model.
5. Classifier Evaluation.
a)Whats the error rate of a classifier?
b)Is the error rate on old data likely a good indicator of the error rate on new data? Why?
c)To predict the performance (error rate) of classifier on new data. What do you need to do?
6. Error Estimates.
a)Use an example to describe what stratification is
b)Use an example to describe what N-fold cross validation is
c)Whats the standard way of predicting the error rate of learning technique, given a single , fixed sample of data? Whats the standard procedure of using it?
d)What are the two most popular ways of measuring the error rate of a learning scheme?
7. Association Rules.
a)Describe how associate rules find other most likely interested items when an item is purchased.
b)Describe how associate rules find other most likely interested items when two items are purchased.
8. Artificial Neural Net (ANN)
a. Give an example of the sigmoid functions used in ANN.
b. Use perceptions to create the following logic gates: 1) AND 2) OR 3) NOT 4) XOR
c. Draw ANN with 1 input layer (with 2 inputs), 1 hidden layer (with 3 nodes) and 1 output layer with two outputs.
d. Whats the minimal number of hidden layers needed to approximate any continuous multivariate function to any desired degree of accuracy?
9. Genetic programming?
a. Describe what genetic programming is.
b. How does genetic programming create the mutation.
c. How does genetic programming create the cross-over.
d. How does genetic programming create offsprings.
e. Use an example to show what a terminal set is. f. Use an example to show what a function set is.
10. Page Rank?
a. Describe how the page rank algorithm works.
b. Markov chain convergence theorem is underlining the math that makes page rank work. Whats Markov Chain convergence theorem.
c. Find (lim n -> infinity P power N), where the 2*2 matrix P = 1/3 2/3 1/7 6/7
11. Discretization.
a. Describe two methods of unsupervised discretization.
b. Describe a method of supervised discretization.
12. Deep Learning.
a. Describe the auto encoder.
b. Describe the sparse coding.
c. Describe the convolution neural network. d. Whats pooling?
Explanation / Answer
Ans 4) Meta Modelling:
Metamodeling is defined as analysis, creation and development of frames, rules, constraints, models and theories applicable and useful for modeling a set of problems. It defines the languages and processes that are used to form a model.
A metamodel is a model that defines the language for expressing a model.
Stacking is a method in which various models are combined to create a meta learner. Stacking is used to combine models of different types. The Stacking operator is a nested operator. It has two subprocess: the Base Learners and the Stacking Model Learner subprocess.
Related Questions
drjack9650@gmail.com
Navigate
Integrity-first tutoring: explanations and feedback only — we do not complete graded work. Learn more.