Artificial Intelligence Ch 6

Réussis tes devoirs et examens dès maintenant avec Quizwiz!

Logistic regression is: -Neither supervised nor unsupervised. -An unsupervised machine learning algorithm -A supervised machine learning algorithm -Both supervised and unsupervised.

3

What is SRM? -Special Regression Minimization, a method for regression using SVM. -Support Regression Machine, SVM used for regression -Structural Risk Minimization, a modelling concept -Structural Risk Minimization, a type of SVM

3

A Bayesian Network consists of _____. -Nodes and Arcs. -Keys and Edges. -Nodes and Edges. -Keys and Arcs.

1

Complete the sentence: SVM model is based on (1)_____ the data and constructing (2)_____ between the categories obtained. -(1) mapping; (2) hyperplanes -(1) mapping; (2) parameters -(1) changing; (2) models -(1) predicting; (2) curves

1

How can SVM be classified? -It is a model trained using supervised learning. It can be used for classification and regression. -It is a model trained using unsupervised learning. It can be used for classification and regression. -It is a model trained using unsupervised learning. It can be used for classification but not for regression. -It is a model trained using supervised learning. It can be used for classification and not for regression.

1

Let A : It will rain on Monday. If P(A) = 1, then _____ -It will definitely rain on Monday. -It may rain on Monday. -Not enough data provided. -It will definitely not rain on Monday.

1

Let's say you want to group similar food items from your grocery cart into groups of canned goods, produce, and meats while unloading them onto the checkout counter. You keep doing this process until you run out of items. What is this an example of? -K-means clustering -None of these are correct -Association -Supervised learning

1

What is a limitation of machine learning? -Too much entropy, and a machine might get bogged down in the volume of data. -It may teach us something new. -It can recognize patterns and draw conclusions where a person might not. -It can process large amounts of data in short periods of time.

1

What is entropy? -A measure of the randomness in the information being processed. -A conclusion drawn from a set of information. -A relationship determined from a set of information. -None of the other answers.

1

What is the algorithm that goes through the process of sorting clusters based on similarities and averages them into their own centroid? -K-means clustering -Clustering -Association -Machine learning

1

What is the purpose of the hyperparameter C in the Soft Margin Classifier? -It dictates how 'soft' the model is. -It dictates how linearly separable the model is. -It dictates how good a model can classify data. -It dictates how good a model can predict data.

1

Which of the following are regression problems? 1. Predict the water temperature based on salinity. 2. Predict the maximum temperature given the minimum temperature. -1 and 2 -1 -2 -Neither of these

1

Which of the following is correct? -Bayesian network is a directed acyclic graph. -Bayesian network is a directed cyclic graph -Bayesian network is a undirected cyclic graph. -Bayesian network is a undirected acyclic graph.

1

Which of the following is extensively used in Speech Recognition Systems? -Trigram Statistical Language Model. -None of these. -Bayesian Networks. -Digital-to-Analog Converter.

1

Which of the following is used to extract information from a sound wave in Speech Recognition Systems? -MFCC -ML -BNN -PHP

1

Which of the following nodes form a decision network when combined with a Bayes belief network ? -Utility node and decision node -Chance nodes only -Chance nodes and decision nodes -Utility node and chance node

1

You have trained a computer vision model to recognize pictures of cats. It works very well except when shown a hairless cat, which it does not classify at all. What might be the problem? -The training data did not feature any pictures of hairless cats. -The computer vision model is flawed. -Poor image quality. -The training data did not have enough cat images.

1

Deep learning allows dealing with much more information than other approaches. Mark the option that makes it interesting for NLP applications. -Machine learning cannot be applied to NLP. -Neural language processing needs lots of data. -Texts may be represented by forms more complex than simple bag of words. -The traditional machine learning task regression is no longer necessary.

3

Given, P(¬A) = 0.7 P(A∧B) = 0.02 P(A∨B) = 0.3 P(B) = _____ -0.14 -0.02 -0.2 -1.2

2

High entropy is an indication of: -That there is an observable relationship in the information -The randomness in the information being processed -How easy it is to draw conclusions from the information -Predicting coin toss results

2

Mark the only sentence that is true. -Shallow learning is the simplest form of deep learning. -Deep learning is a subdivision of machine learning. -Machine learning is a subdivision of deep learning. -NLP is the acronym for Neural Language Processing.

2

The cost function that is used in logistic regression is: -Both of these -Maximum Likelihood -None of these -Mean Squared Error

2

We cannot apply a logistic regression algorithm on: -A 3-class classification problem. -A regression problem. -A 2-class classification problem. -A multi-class classification problem.

2

What is a neural network? -The neurons in the human brain. -The layers of a deep learning model which take input and pass output to the next layer. -A network of computers that communicate and share data to work on a problem. -A collection of computing nodes, each working on a single task.

2

What is image classification in AI? -How useful an image might be to a computer vision AI. -The task of making predictions about what is represented in an image. -Identifying whether or not some data contains an image. -Classifying an image based on its' quality.

2

What is one of the main drawbacks of using the unsupervised learning method for all situations? -It draws too much power -The outcome is unknown -The cost is too great -It's not as efficient

2

What is the purpose of the Kernel Trick? -To transform the problem from nonlinear to linear -To transform the data from nonlinearly separable to linearly separable -To transform the problem from regression to classification -To transform the problem from supervised to unsupervised learning.

2

Which of the following statements about decision nodes is false? -They represent decision variables and have a utility value for each decision. -Decision nodes have utility nodes as parents. -The information available to them while taking decisions is represented by edges coming into them. -In sequential decision making problems, where multiple decision variables are involved, the nodes are ordered entirely.

2

Consider the logistic regression model. What is the range of the sigmoid function? -(-inf, 0 ) -(0, inf) -(0, 1) -(-inf, inf)

3

Given, P(A|B) = 0.4, P(A) = 0.6 and P(B) = 0.2. Find P(B|A). -0.096 -0.0266 -0.133 -0.048

3

What is the purpose of machine learning? -Pattern recognition, computation, and predicting outcomes -Accurately drawing conclusions that were not previously known -All of the answers are correct -Dealing with large sets of data

3

Which are the three steps of deep learning? -Gather a training set, build a deeper set, and generate a random forests learner. -Gather a testing set, build a deeper, and use neural networks. -Gather a training set, build a model using neural networks, apply the model to a testing set. -Gather a testing set, build a model using random forests, generate a learner.

3

Which of the following concepts is used to convert the extracted information from a sound wave to text transcriptions in Speech Recognition Systems? -Bayesian Networks. -Data Mining. -Pattern Matching. -Data Science.

3

Which of the following is kept at each node in an Bayesian Network? -Conditional Probabilities. -Binomially Distributed Probability Values. -Local Probability Table. -Normally Distributed Probability Values.

3

Which of the following is used to calculate probability of the nodes in a Bayesian Network if conditional probabilities are present? -None of these. -Normal Probability Distribution Theorem. -Bayes' Theorem. -Binomial Theorem.

3

Why is machine learning important? -It is branch of computer science that deals with pattern recognition, computation, and predicting outcomes. -None of the answers are correct. -It can tell us something new about the information that we may not be able to see ourselves. -It can deal with the same information a person can.

3

Bayesian Network consists of: -Normally distributed Probabilities. -FOLs. -Binomially distributed Probabilities. -Probability table.

4

Decision networks are _____ -Not used for decision making in complex, dynamically changing scenarios. -Not suitable for making sequential decisions. -Special cases of Bayes Belief Networks with additional nodes for decision and utility. -Generalized versions of Bayes Networks with additional nodes for decision and utility.

4

Expected utility of an agent's action/decision is _____ -Value of agent in the decision making process -Sum of probabilities of all random variables in the decision problem -Sum of probabilities of all random variables multiplied by value of agent -Sum of probabilities of decision/action multiplied by utility value of that decision/action

4

Given, P(A) = 0.2 P(B) = 0.3 P(A|B) = 0.4 P(B|A) = _____ -0.5 -0.06 -0.3 -0.6

4

Given, P(¬S) = 0.6. P(S) = _____ ? -0.2 -0.3 -0.7 -0.4

4

What is a distinguishing feature of deep learning? -Creating algorithms that allow decisions to be made about various inputs. -The ability to analyze information and make a prediction. -Training a model to recognize patterns. -Passing data through layers of transformation and analysis to make a prediction.

4

Which of the following is true? -Language Models are also known as Probabilistic Language Models. -None of these. -Language Models are used to provide text transcriptions for different languages. -Language Models use Probability to predict the next word.

4

Which of the following machine learning algorithms is the base for deep learning? -Boosting -Support Vector Machines -Random Forests -Neural Networks

4

Which of the following mathematical tools can be used to convert time domain to frequency domain in a sound wave? -Z Transforms. -None of these. -Laplace Transforms. -Fourier Transforms.

4

Which of the options is not a NLP application? -Named entity recognition -Translation from English to French -Sentiment analysis -Translation from C++ to Java

4

While evaluating a decision network in a sequential decision problem, an agent has to look for piece of information that might or might not help in making an optimal decision. The price to be paid/lost time in seeking the information is called _____ -Value of perfect information -Maximum expected utility of the information -Utility of information -Value of information

4

You are working with an image classification AI and need it to identify whether an image features a dog. What is the important step the AI needs before it can function? -Plug in a webcam so it can see the world. -Train the AI with pictures of various animals. -Take a picture of a dog and try to classify the image. -Train the AI with pictures of dogs.

4

_____ is a learning model that is used to identify a relationship between large amounts of information from a data set. -Unsupervised learning -Clustering -K-means clustering -Association

4

__________ is an algorithm used in machine learning that groups items of similarity without using labels. It can sometimes be known as the 'true artificial intelligence' algorithm. -Supervised learning -All of these are correct -Association -Unsupervised learning

4


Ensembles d'études connexes

Business Innovation Final Set #2

View Set

Chapter 01 - A Preview of Cell Biology

View Set

Physical Science (PSY151) Chapter 5: Les Thomas

View Set

Econ Unit 6 Fiscal Policy: Government Taxation and Spending

View Set

chapter 11 sensation and perception

View Set

GEOG 145 Slavery and Human Trafficking Final Review

View Set

Principles of Macroeconomics ASU-BEEBE Final

View Set