Artificial Intelligence Final Exam

An Artificial Neural Network (ANN) is a collection of Neural Networks (NNs).

False

Frank Rosenblatt created the perceptron in 1985.

False

Google Duplex is good at translating text between languages, but struggles at tasks like scheduling a haircut appointment.

False

In Connect Four, users take turns dropping different colored checkings into a 3-by-3 grid.

False

In K-Nearest Neighbors, the algorithm will always find the 5 closest points to a given data point.

False

In the Last Coin Standing game, the winner is the player who removes the last coin from the pile.

False

In the hexapawn game, only pawns and bishops are used.

False

In the minimax algorithm, the first half (top of the tree) of the levels maximize value choices and the second half (bottom of the tree) of the levels minimize the choices.

False

In the movie recommendation system described in chapter 8, you pass the system a genre and it recommends 3 movies that fall into that genre.

False

It's possible to integrate a chatbot into a web page, but it's not possible to connect your chatbot to popular messaging platforms like Google Assistant, Facebook Messenger, Kik, Viber, etc.

False

The use of chatbots, the cloud, and voice biometrics, companies can improve their customer service and handle more calls with less staff.

True

To compare the similarity of two data points, you can use the Euclidean or Pearson data score.

True

What is the name of the Python module that is most commonly used for evolutionary computing?

deap

Which of the following is a part of the backpropagation algorithm?

partial derivatives

The logistic function used by neural networks to "squish" the data into the [0..1] range is called the _____ function.

sigmoid

In the first neural network video, the example consisted of _____ output layer(s).

10

The goal for learning in the first neural network video was to optimize _____ weights and bias values.

13,002

To open a chess game, the first player has _____ unique moves to choose from.

20

In the first neural network video, each image consisted of _____ pixels.

784

The human brain has approximately _____ neurons.

85 billion

The Brown Corpus is named after _____.

Brown University

The biggest drawback of the easyAI Python module is that you cannot play human vs. human games.

False

The global market for chatbots is expected to shrink rapidly from now until to 2024.

False

When designing your own chatbot, you should not focus on a well-defined and narrow domain. It is best to create a broad, general "jack of all trades" chatbot.

False

The images in the example shown in the video came from the _____ dataset.

MNIST

NLTK stands for _____.

Natural Language Tool Kit

Which of the following is not normally used as a way to terminate the main loop of an evolutionary program?

Network congestion

To solve the N-city Traveling Salesman Problem using exhaustive search, it would take _____ time.

O(N!)

What did the robot controller try to consume?

Pound signs

_____ is the process of identifying parts of an image.

Segmentation

____ is a technique that converts text into a numerical form so that you can use machine learning on it.

The Bag of Words Model

A neural network is a collection of connected neurons.

True

An Agent is a system that can handle all the conversations and route all the necessary actions. It is a natural language module that gets trained frequently to cater to use-specific requirements.

True

Many people use smart home assistants like Alexa, Siri, and Google Home for tasks such as: Listen to music Set alarms Create shopping lists Get weather forecasts Control devices around the house Order online commerce goods Book a plane ticket

True

Minimax is an algorithm often used for making decisions in two-player games

True

The K-Nearest Neighbor Classifier will predict the output for a data point based on how many neighbors belong to each class.

True

The context of a chatbot is used to give coherence and fluency to a discussion, preserving the key concepts that have already been used in the conversation.

True

The easyAI Python library provides the AI framework for two-player games.

True

Using similarity scores, collaborative filtering is used to find the K most similar objects to a given data point.

True

Warren McCulloch and Walter Pitts created a computational model for neural networks in 1943.

True

When creating a Python game using the easyAI module, you must define methods possible_moves, make_move, scoring, show, and is_over.

True

When using minimax, an evaluation function is necessary to compute the "goodness" of the state of the game.

True

What would be the output of the following program? a = [1,2,3,4]b = ["Q","U","I","Z"]print(list(zip(a,b)))

[(1, 'Q'), (2, 'U'), (3, 'I'), (4, 'Z')]

What would be the output of the following program? def thing(x):return 2*x+3 y = [0,2,4,6]print(list(map(thing,y)))

[3, 7, 11, 15]

The learning process of applying small "nudges" to the weights and biases of a neural network is called _____.

backpropagation

A _____ will divide text into pieces. Each piece may have a couple hundred words in it.

chunker

Naive Bayes was used in several programs last week for _____.

classification

Which of the following is not one of the main components in evolutionary computing?

deterministic algorithm

Google provides _____, which is a platform to easily create text- or voice-based bots.

dialogFlow

Which technique was used by the robot controller program?

evolutionary programming

Which technique was used by the bit counter program?

genetic algorithms

The learning technique in which nudges of weights and biases are made to reduce the value of the cost function is called _____.

gradient descent

A ____ follows the process of grouping together the different inflected forms of a word so they can be analyzed as a single item.

lemmatizer

Which of the following is not considered an evolutionary computation paradigm?

neural networks

A sentiment analyzer can determine if text is _____.

positive or negative

What color is the 3-D map that models the space described in this problem?

red

A _____ produces morphological variants of a root/base word. The variants may not be actual words.

stemmer

Instead of adjusting weights and biases after every data point, random samples of mini-batches are used. This is called _____.

stochastic gradient descent

Which of the following is not a common intent for a chatbot?

swindle

When multiplying the 3 ratios in the backpropagation technique, _____ is used.

the chain rule

The cost function for the digit classification program is ____.

the sum of the squares of the difference between actual values and predicted values

Which of the following is not a type of layer in a neural network?

threshold

A _____ breaks text into a set of pieces, such as words or sentences

tokenizer

Latent Dirichlet Allocation is a _____.

topic modeling technique

Which of the following is not a part of a human neuron?

weight