ACC 3510 Chapter 12
four
If there are two possible classifications, then the confusion matrix would show four cells with the prediction options.
mathematical
In a machine learning context, a model is a mathematical representation of some process.
True
True or false: AI applications are also called cognitive technologies.
False
True or false: Classification problems seek to minimize the differences between predicted and actual values.
FALSE
True or false: Since blockchains are immutable, auditors do not need to audit blockchain transactions.
False
True or false: Supervised learning is a type of machine learning but it is seldom used in practice.
Is it different?
Anomaly detection algorithms address which of the following questions? Is it different? Is it A or B? What should I do next? How much is the predicted price?
engine
Artificial neural networks are the engine of machine learning.
two
Deep learning networks have more than two non-output layers.
A true positive is a correct classification of a spam email message; a false negative is an incorrect classification of a spam email message
In an example where the goal is to identify spam email, which of the following best describes the difference between a true positive (TP) and a false negative (FN) in a confusion matrix? A true positive is the incorrect classification of a good email; a false negative is the incorrect classification of a spam email A true positive is a correct classification of a spam email message; a false negative is an incorrect classification of a spam email message A true positive is the correct classification of a good email; a false negative is the incorrect classification of a good email A true positive is a correct classification of a spam email; a false negative is a correct classification of a good email
training
In machine learning, the model learns from training cases or data.
TRIAL and error
In reinforcement learning, the model learns by TRIAL and error.
output
In supervised learning, the output is a known set of values that the neural network seeks to predict.
similar
In unsupervised learning, the model discerns how elements of the dataset are similar or not
patterns
In unsupervised learning, the system uses unstructured data and the system discovers patterns in the data.
distributed and decentralized, consensus, and immutability
List the key characteristics of blockchain technology: distributed and decentralized, consensus, and immutability
public, private, and consortium
List the three types of blockchain public, private, and consortium blockchain.
data.
Machine learning applications improve their ability to analyze patters as they process more data.
ARTIFICIAL
Machine learning is a subset of ARTIFICIAL intelligence.
Match the key concepts behind Bitcoin versus Ethereum.
Match the key concepts behind Bitcoin versus Ethereum. Bitcoin: Anonymous peer-to-peer transactions, no middleman involved. Ethereum: Uses smart contracts
Match the key concepts behind Bitcoin versus Ethereum.
Match the key concepts behind Bitcoin versus Ethereum. Bitcoin: One block is added to the blockchain every 10 minutes. Ethereum: One block is added to the blockchain every 12 to 15 seconds.
classification
Most machine learning applications are designed to perform either some sort of classification or regression.
Language
Natural language processing (NLP) applications communicate using written and spoken Language
models or algorithms
Neural networks are mathematical models that convert inputs to outputs/predictions.
repetitive
RPA is a tool that can perform high-volume, repetitive, tasks such as preparing tax returns or managing accounts payable.
PREDICT
Regression problems seek to PREDICT real numbers, such as house prices.
human
Robotic process automation is used to reduce the human labor required.
Satoshi Nakamoto
Satoshi Nakamoto published a white paper in 2008 to introduced a concept on distributed ledger system which is the foundation of the blockchain technology.
All transactions on a blockchain network must be verified by a central authority.
Select a wrong statement about blockchain concepts. All transactions on a blockchain network must be verified by a central authority. Participants on a blockchain network must reach consensus before transactions are committed to the blockchain. Past information recorded on a blockchain cannot be edited or altered. Blockchain technology uses a distributed ledger system and each copy of the ledger containing the same transaction records.
Miners are the nodes/machines creating and validating blocks.
Select the best statement in describing the concept of miners in blockchain. Miners can use various consensus algorithms while creating blocks on a blockchain network. Miners are any nodes/machines in a blockchain network. Miners are the nodes/machines creating and validating blocks.
Private blockchain is also called enterprise blockchain.
Select the correct statement on the three types of blockchain. A public blockchain requires permission to join. Consortium blockchain is permissionless. Private blockchain is also called enterprise blockchain.
confusion matrix.
The success of an artificial neural network can be measures by using a confusion matrix.
A smart contract is written in software code. A smart contract defines the terms and business rules to be used in conducting transactions. A smart contract defines the digital assets that could be transferred in transactions.
What is a smart contract used in a blockchain network? A smart contact defines the transaction rules in Bitcoin. A smart contract is written in software code. A smart contract defines the terms and business rules to be used in conducting transactions. A smart contract defines the digital assets that could be transferred in transactions.
To eliminate intermediaries/middlemen in perform transactions
What is the original purpose of using a distributed ledger system with the blockchain technology? To simplify accounting concepts involved in conducting transactions To eliminate intermediaries/middlemen in perform transactions To reduce transaction cost To reduce transaction time.
What action should I take next?
What kinds of business questions would a reinforcement learning algorithm help answer? Will a 20% discount attract buyers? What is this house price? What action should I take next? Is this website safe?
Recognize faces
Which of the following are NOT uses of natural language processing and understanding applications? Discern sentiment Translate text Recognize faces Extract semantic meaning
Incentive-based learning
Which of the following is NOT a technique to train a neural network system? Incentive-based learning Supervised learning Unsupervised learning Reinforcement learning
Information moves from the output layer to the input layer
Which of the following is NOT true about artificial neural networks? They can be nested so the overall network includes multiple layers Connections between nodes can include loops Information moves from the output layer to the input layer Information moves through hidden layers
A neural network with more than 100 layers cannot be effectively trained
Which of the following is NOT true about modern neural networks? Neural networks have been used to solve sophisticated problems like computer vision. More layers increasing the complexity of the models A neural network with more than 100 layers cannot be effectively trained Most business problems require only two or three hidden layers
Deep learning involves less complexity than machine learning
Which of the following is NOT true about the relationship between AI, machine learning, and deep learning. Machine learning encompasses deep learning Deep learning involves more complexity than machine learning Artificial intelligence includes both machine and deep learning Deep learning involves less complexity than machine learning