Dijkstra’s Algorithm

• Finds the shortest path between two points on a graph

• Each point on the graph is called a node or a vertex

• Steps:
  • Give the start vertex a final value of 0
  • Give each vertex connected to the last vertex a working value, if it already has a working value, only replace it with another if that’s lower
  • Check the working value of any vertex that has not been assigned a final value, assign the smallest value to that vertex, that is now its final value
  • Repeat steps 2 and 3 until the end vertex is reached, and all vertices have been assigned a final value
  • Trace the route back from the end vertex to the start vertex to find the shortest path between those two vertices

A* algorithm

• Based on Dijkstra, but adds an extra heuristic value — a guess on how far we have to go to reach the next destination most efficiently

• Examples on a heuristic value is the diagonal remaining distance of the node to the end vertex

• For example, the two values are added and the path with the smallest sum is the chosen route

Machine learning categories

• Supervised learning
  • Known input and output data is made available
  • Learning will end when the model produces an acceptable level of performance
  • Regression-type: target will be to predict a quantitative value such as a number
  • Classification-type: target is a qualitative value such as whether a customer is worth sending marketing emails to

• Unsupervised learning
  • Only a measured set of input values but no corresponding output
  • The task is to identify some underlying structure or the distribution of the dataset

How artificial neural networks have helped with machine learning

• Deep Learning
  • Artificial neural networks are based on the interconnections between neurons in the human brain
  • Handle huge amounts of data

• Machine Learning
  • Systems that learn without being programmed to learn

• Reinforcement Learning
  • Learning on the basis of “reward and punishment”
  • Uses trial and error in algorithms to determine which action gives the most optimal outcome

Back propagation of errors and regression methods

• Back propagation
  • The initial outputs from the system are compared to the expected outputs and the system weightings are adjusted to minimise the difference between actual and expected results.
  • Calculus is used to find the error gradient in the obtained outputs: the results are fed back into the neural networks and the weightings on each neuron are adjusted (note: this can be used in both supervised and unsupervised networks).
  • Once the errors in the output have been eliminated (or reduced to acceptable limits) the neural network is functioning correctly and the model has been successfully set up.
  • If the errors are still too large, the weightings are altered – the process continues until satisfactory outputs are produced.

• Regression
  • To make predictions from given data by learning some relationship between the input and the output