Artificial intelligence and machine learning Flashcards
Describe what Artificial Intelligence is
Artificial intelligence: The field in which machines are programmed to mimic human intelligence.
Describe what Machine Learning is
Machine learning (ML): A subset of AI that involves the use of data and algorithms to imitate the way humans learn, gradually improving its accuracy. Machines look at the patterns in data provided and begin learning from those patterns to make better predictions and decisions in the future.
Describe AI and machine learning and how they are different
Artificial intelligence is the field, machine learning is a subset of AI which focuses on applying specific algorithms that allow the computer to learn information without being told how to do so and improve their performance on a task through experience. It focuses on developing systems that can learn from and make decisions based on data.
State the types of machine learning
Supervised learning
Unsupervised learning
semi-supervised learning
Describe supervised learning
Supervised learning: the algorithm is trained on a labelled data set (the input comes with the correct output). Examples include classification and regression tasks.
* Labelled datasets
* Designed to train or “supervise” algorithms into classifying data or predicting outcomes.
* Both the input data and the target data are provided therefore we are helping the algorithm to map directly between the input (e.g., HbA1c) and the target (diabetes).
* Supervised learning takes a long time to prepare and often require multiple experts to rate each sample.
Describe unsupervised learning
Unsupervised learning: the algorithm is given data without explicit instructions on what to do with it. It must find patterns and relationships within the data. Examples include clustering and association tasks.
- There is input data but no target (unlabelled)
- The intention here is to allow the algorithm to create its own separation and learn to classify the target
- These are often much more sophisticated algorithms, but they are more prone to training failures and classification errors
- Unsupervised learning can be very useful if we have a huge amount of information but aren’t sure what the exact answer is that we are looking for
Describe semi-supervised learning
semi-supervised learning:
* Both labelled and unlabelled data
* ideal for medical image data e.g., a radiologist can label a small subset for MRI scans for tumours so the machine can grade tumour severity
How can supervised learning be implemented
Supervised learning for classification and prediction
- Classification – assign data into specific categories for e.g., support vector machines, decision trees and random forest
Binary classification: Predict one of two classes (can all be expressed as 1 or 0) e.g. Diabetes or no diabetes.
Multi-classification: Predicting the presence of more than 1 class in some dataset.
E.g. MRI training data and brain tumour targets.
The system would learn to predict anything from 0 to 3 classes here depending on the presence of tumour and contained sub-tissues.
Give examples of Supervised learning for classification and prediction
Support vector machine and decision tree
Decision tree: a supervised learning algorithm used for both classification and regression tasks. It models decisions and their possible consequences in a tree-like structure where:
What is a decision tree
Decision tree: a supervised learning algorithm used for both classification and regression tasks. It models decisions and their possible consequences in a tree-like structure where:
o Nodes represent the features (attributes) of the dataset.
o Branches represent the decision rules.
o Leaves represent the outcomes (either a class label in classification or a continuous value in regression).
- makes yes/no decisions about each feature until a target prediction is reached.
- During the training stage the questions asked at each decision node are tuned to an optimal threshold that gives a good result.
- The intuition behind this can be compared to human decision making. For example, if determining whether to go for a walk, we might ask ourselves questions on the weather, whether it’s cold, whether we’re tired, if we need to priorities something over walking…
Decision trees are a good starting point in machine learning as we can visualize their structures and understand how they have solved a problem.
How can regression be implemented
Regression – uses an algorithm to understand the relationship between dependent and independent variables for e.g., linear regression, logistic regression and polynomial regression
How can unsupervised learning be implemented
To analyse and cluster unlabelled data sets. To discover hidden pattern in the data without the need for human intervention
Give examples of unsupervised learning
- Clustering – grouping unlabelled data based on their similarities or differences for e.g., K-mean clustering
- Association – for finding relationships between variables in a dataset
- Dimensionality reduction - a learning technique used when there are a large number of variables (or dimensions) in a given dataset. It reduced the number of variables while persevering the data integrity for e.g., Principal component analysis which can be used in the pre-processing data stage
Discuss the advantages of Artificial Intelligence
Efficiency: require few computational resources to work on real life data.
Time and cost saving
Can analyse data from multiple sources
Error detection: no human error (error reception/accuracy)
Productivity: smaller jobs can be taken from humans and given to AI
Transferability: can operate across a wide variety of problems and industries.
Discuss the disadvantages of Artificial Intelligence
Redundancy: removing humans from certain workforces (job displacement)
Datasets: need to provide AI with data, can be time consuming and expensive.
Understanding: users willingly handing over information (privacy)
