Professional Documents
Culture Documents
Supervised, Unsupervised & Reinforcement Learning
Supervised, Unsupervised & Reinforcement Learning
Supervised, Unsupervised & Reinforcement Learning
&
Reinforcement Learning
Supervised Learning
This is where you feed the machine learning algorithm labelled training data. The
labels contain the expected outcome for that data. The machine used the labels
and training data to train the model.
Labelled data is split into training and test data
Not - Spam Examples
Spam Examples
Spam Spam
Training data used to train the model
Training Data
SPAM Classifier
Not - Spam
Model
Spam
(e.g. Naive Bayes)
Trained model is ready to test
SPAM Classifier
Model
Spam
SPAM Classifier
Model
97%
Accuracy
The tested model is then ready to deploy
SPAM Classifier
Model
✔️
(e.g. Naive Bayes)
The trained model can then be deployed
SPAM Classifier
Model
✔️
(e.g. Naive Bayes)
Classification vs Linear Regression
Classification Regression
The aim of the algorithm is to split data into The aim of the algorithm is to predict the value
predefined categories. of a dependent variable based upon another
explanatory variable. The example shows
simple linear regression.
Unsupervised Learning
Machine learning algorithm is trained on unlabelled data and is left to cluster data
itself. Certain hyper-parameters may be set (such as how many clusters to form)
but the process is generally unstructured.
● Useful for categorising many different objects
● Identifying hidden trends or patterns
● Anomaly detection(e.g. fraudulent transactions, spotting skin cancer, crime
detection)
Clustering Algorithms
Clus
The aim of a clustering algorithm is to pick up te r
patterns that allow input data that can be
separated into groups.