Machine Learning Lecture

Machine Learning Defs and notes

Supervised Learning

A machine learning algorithm that learns to make predictions by training a model on a labelled dataset.

Unsupervised learning

A machine learning algorithm that finds patterns in a dataset with no given labels

Reinforcement Learning

A machine learning algorithm where an agent interacts with an
environment and learns to take actions that maximise a reward

K-means algorithm

Unsupervised machine-learning algorithm that can be used for clustering. In a K-means approach, data is grouped in clusters based on similarity. To do this, we define the number of clusters (K) and then K centroids are selected randomly. For each sample in the dataset, the closest centroid is found, and the sample is assigned to that centroid. The centroids are recalculated based on the samples of the cluster, and the process is repeated until a specific number of iterations is reached or there is no change in the centroids. Finally, the data is divided according to the closed centroid.

Machine Learning

Subset of artificial intelligence (AI) methods that allow

computers to learn from and make predictions or decisions

based on data .

Classification

Assigns input data to a label based on patters or characteristics. Examples: Spam detection and handwriting recognition.

Regression

Models the relationship between a dependent variable and one or more independent variables to make predictions. Examples being: Energy consumption prediction and house price prediction.

Gradient Descent

Iterative optimisation to find the minimum of a function (cost function/loss function) by adjusting the model parameters in the direction of the steepest descend. The idea is to update the parameters in the direction that minimises the cost function.

Decision trees

A supervised machine learning algorithm used for classification or regression tasks. Makes decisions based on the input data to provide a prediction and by asking a series of questions based on the features of the input data to output a prediction.

Data

The information that we use to build, train and evaluate machine learning models. There are two types, categorical: text and numerical: numbers. After training, it must be validated by using a different subset of data to make a prediction that validates the trained model. If unsuccessful, model is trained again until satisfactory. After validation, the model goes through testing(data, trained model, prediction, evaluation). Training data must be varied for less error.

Confusion Matrix

A confusion matrix is a method to summarise the performance of a classification algorithm using a table (matrix).