Machine Learning

Machine Learning (ML) in IoT:
  • Is used to extract knowledge from IoT data
  • Extraction can take place at the edges or in the cloud
  • Popular algorithms: supervised learning, unsupervised learning and reinforcement learning
  • Examples of machine learning systems: Amazon Machine Learning, Google's TensorFlow

Unsupervised Learning:
  • For input data that do not have known grouping labels
  • Perform analysis to find hidden patterns or groupings (clusters) in input data based on similarities
  • Similarity measures are defined based on Euclidean or probabilistic distance
  • Common clustering algorithms are hierarchical clustering, k-Means clustering, self-organizing maps

Supervised Learning:
  • Uses a known dataset (training dataset) with known labels to build models
  • Models work with new/unseen input dataset
  • Supervised learning algorithms include Classification (such as decision trees) and Regression (such as linear/nonlinear regression)

Reinforcement Learning:
  • Learn by interacting with an environment
  • Agent learns from the consequences of its actions
  • Agent selects its next actions based on its past experiences (exploitation) and on new choices (exploration)
  • Agent receives a reward which indicates the success of an action's outcome
  • Agent strives to select actions that maximize the accumulated reward over time