Artificial Intelligence LinkedIn Course

the network is making predictions that are turning out to be very wrong.

A scientist says that the back propagation of errors is correcting for guesses that have steep gradient descent. What is that saying about the network?

classified data works with human created classification or labels

Difference between classified and clustered data

Naive Bayes is naive because it can classify even when predictors are autocorrelated

Imagine that you work for a university that wants to use machine learning and Naive Bayes to predict which students might have difficulty graduating. You create 3 predictors; financial hardship, grade point average, and class attendance. In a meeting, a data scientist points out that you might not want to use attendance and grade point average because they are strongly autocorrelated. How might you answer this question?

Training data is the data the machine uses to learn, test data is a sample of data to test what it learned.

In supervised machine learning what is the difference between training data and test data?

this deep learning artificial neural network can identify very complex patterns

Key advantage of having several hidden layers in artificial neural network

Humans tend to describe things as qualities which are difficult to translate into search results.

One of the biggest challenges for AI natural language processing

Unsupervised Learning

Preschoolers grouping blocks into groups which make sense to them (i.e. by color or letter on block) is what type of learning?

If the company's data has set rules and patterns

What determines whether a company is or is not a good candidate for AI?

there is a higher price to pay if the robotics make a mistake

What financial reason would cause a robotics company to use AI rather than machine learning?

A symbolic system needs to be programmed to connect symbols to patterns, while machine learning discovers patterns by looking at the data.

What is the biggest difference between symbolic and machine learning?

finding patterns in symbols is not the same as intelligence

What was the core finding of the Chinese room experiment?

Because these systems have benefited from access to massive amounts of digital data.

Why has machine learning become the dominant form of artificial intelligence?

They are tools to help you decide what reports you'd like to see or problems you'd like to solve

Why is it important to understand different learning algorithms?

Because the bias is a reaction to the adjustments in the weights of the connections

With an artificial neural network, the weights are added to the connections, but the bias is added to the neuron. Why does the network need to be set up this way?

The local bus has a high variance, and the express bus has a high bias

You take a new job where you need to arrive by bus in 30 minutes. There are two buses available to you. The first is a bus that makes all the local stops. This local bus is quicker if there are just a few people. The second is an express bus that drops you off a few blocks from your job. You try both options the first week. The local bus gets you to work in 20minutes, 30 minutes, and 45minutes. The express bus always gets you to work in 35 minutes. What is the challenge with these two options?

research the character traits of the cluster with the highest number of diabetics

Your insurance company covers a lot of people with diabetes. The company wants to research different characteristics of people with this disease so that they can intervene and try to change customer behavior to prevent the disease. How can your insurance company use K means clustering to help?