18.2 Artificial intelligence, machine learning and deep learning

Describe the purpose of both the A* algorithm and Dijkstra’s algorithm.
- to find the optimal / shortest / most cost-effective route
- between two nodes in a
- based on distance / cost / time.

Describe the purpose of a graph when used in an Artificial Intelligence (AI) system.
- A graph is used in AI to record relationships between entities
- … using vertices / nodes and edges
- for example, to represent places on a map and the distances between them, in order to find the shortest route.

Explain what is meant by Deep Learning in relation to Artificial Intelligence (AI).
- Deep learning learns by finding hidden patterns that are undetectable to humans.
- It structures algorithms in layers: input layer, hidden layers and output layer.
- ….. to create an artificial neural network to learn and make intelligent decisions on its own.
- It is trained using large quantities of unlabelled data.
- Deep learning requires/uses a large number of hidden layers.
- ….. the larger the number of layers, the higher the level of success.

Explain the use of artificial neural networks in Deep Learning.
- Artificial neural networks are designed to work in the same way as the human brain
- ANNs provide the architecture and algorithms for learning from the data
- They have a large number of connected processing units / nodes
- … that are arranged in layers / interconnected and work together to process data
- Deep learning models learn from data by adjusting the weights/biases of the connections between neurons
- They use multiple hidden layers to extract complex features and to make predictions

Describe the structure of a graph as used in an Artificial Intelligence (AI) system.
- A graph uses vertices/nodes to identify/represent entities such as destinations, people, etc
- Edges are used to connect nodes and can represent possible paths between them // a path is the list of nodes connected by edges between two given nodes
- Nodes/edges can be labelled/weighted, and this is a weighting that can be applied and used in the context of the application
- A cycle is a list of nodes that return to the same node.

Explain the use of graphs to aid Artificial Intelligence (AI).
 Artificial Neural Networks can be represented using graphs
• Graphs provide structures for relationships // graphs provide relationships between nodes
• AI problems can be defined/solved as finding a path in a graph
• Graphs may be analysed/ingested by a range of algorithms
• …e.g. A* / Dijksta’s algorithm
• …used in machine learning.
• Example of method e.g. Back propagation of errors / regression methods

Explain how supervised learning and unsupervised learning differ from each other.
- Supervised learning uses labelled data // Unsupervised learning makes use of unlabelled data.
- Labelled data means that known outcomes are applied to specific inputs to help the AI predict outcomes.
- Supervised learning requires initial human input/training // Unsupervised learning does not require human input/training.
- With unlabelled data in unsupervised learning, outcomes are not known
- … the AI has to search for hidden patterns/structures/clusters
- … within the data in order to predict outcomes.

how unsupervised learning takes place in machine learning
- Unsupervised learning uses algorithms to analyse / cluster
- … unlabelled data sets
- They discover hidden patterns / data groupings / clusters without the need for human intervention.
- It is able to discover similarities and differences in data / information.

Describe supervised learning and unsupervised learning.

Supervised learning
- Supervised learning allows data to be collected, or a data output produced, from the previous experience.
- In supervised learning, known input and associated outputs are given // uses sample data with known outputs (in training) // uses labelled input data.
- Able to predict future outcomes based on past data.

Unsupervised learning
- Unsupervised machine learning helps all kinds of unknown patterns in data to be found.
- Unsupervised learning only requires input data to be given.
- Uses any data // not trained on the right output // uses unlabelled input data.