Concepts:

A data set contains objects that have features- color, text, sounds, etc.

For supervised learning, data sets are divided into a training set(80%) and a test set(20%)

In supervised machine learning, the computer learns from training data, which contains features and labels

It then looks at test data , and makes a prediction as to the label, with a confidence score

Machine Learning for Kids adds machine-learning blocks(powered by IBM-Watson) to Scratch

Improving a machine learning model is a cyclical process- the more training data, the more accurate the classifier

Machine learning can do sentiment analysis, classifying the tone of text

Machine learning models can have issues of bias(derived from unequal training data) and overfitting

Training data can be improved by crowd-sourcing

AI Vocabulary:

• Data Set

• Features

• Labels

• Supervised Learning

• Training data

• Test Set

• Confidence Score

• Machine Learning Model

• Sentiment Analysis

• Bias

• Overfitting

Below: Scratch code for a text emotion(sentiment) classifier

Using Images from Google Images:

Open browser to Google Images

Enter a search criteria + jpg, such as Bananas jpg

Click on a tile image

In the image in the pop-up window, right click on "Copy Image Address"

In ML for Kids bin, click on "www" and paste URL of image