Please enable JavaScript.
Coggle requires JavaScript to display documents.
Artificial Intelligence - Coggle Diagram
Artificial Intelligence
The nearest neighbor classifier
Among the simplest possible classifiers
Defining “nearest”
Euclidean distance
measure the distance between any two items
choose the distance metric on a case-by-case basis
count pixel-by-pixel matches
scaling the images
compare the pixels in the bottom-right corner of each image
predict user behavior
users with similar past behavior tend to have similar future behavior
use information about the genre, the artist, and other metadata
relatively scarce and coarse
Collaborative filtering
predict preferences
“filter”
only recommended content that passes
content that you are like
Regression
almost as simple as the nearest neighbor classifier
Linear regression
Learning
given the inputs and the weight
produce the predicted output
actual output isn’t always fully determined by the input
Visualizing
draw a chart eg: cups of coffee
one variable
number of lines of code written
drinking more coffee results in more lines of code being written
line in the chart represents predicted outcome
the intercept
the coefficient
Example
crime rates
real estate prices
software cost
insurance cost
click rates in online advertising
Predict labels
Logistic regression
linear regression method’s outputs into predictions about labels
give us a measure of uncertainty of the prediction
trick to obtain predictions over more than two possible labels
The types of machine learning
Supervised learning
the answers are in the form of yes/no
binary classification problems
given an input
Unsupervised learning
Data visualization
no labels or correct outputs
Reinforcement learning
used in games where the outcome may be decided only at the end of the game
must operate in an environment
good or bad choices is available with some delay