Today, with the advances in the hardware technologies, it is possible to store, process, and output large amount of data. With the increase in the size of data, explaining it, that is, extracting meaningful information from it, becomes a bottleneck. Machine learning, i.e., the science of extracting useful information from data comes as an aid. Empowered with concepts from mathematics, statistics, and computer science, machine learning is arguably the solution for all of our information extraction problems.

Chapter Contents:

• 3.1 Introduction
• 3.2 Data
• 3.3 Dimension reduction
• 3.3.1 Feature selection
• 3.3.2 Feature extraction
• 3.4 Clustering
• 3.4.1 K-Means clustering
• 3.4.2 Hierarchical clustering
• 3.5 Supervised learning algorithms
• 3.5.1 Simplest algorithm: prior
• 3.5.2 A simple but effective algorithm: nearest neighbor
• 3.5.3 Parametric methods: five shades of complexity
• 3.5.4 Decision trees
• 3.5.5 Kernel machines
• 3.5.5.1 Separable case: optimal separating hyperplane
• 3.5.5.2 Nonseparable case: soft margin hyperplane
• 3.5.5.3 Kernel trick
• 3.5.6 Neural networks
• 3.5.6.1 Neurons (units)
• 3.5.6.2 Models and forward-propagation
• 3.5.6.3 Backward-propagation
• 3.6 Performance assessment and comparison of algorithms
• 3.6.1 Sensitivity analysis
• 3.6.1.1 Local sensitivity analysis
• 3.6.1.2 Global sensitivity analysis
• 3.6.2 Resampling
• 3.6.2.1 K-Fold cross-validation
• 3.6.2.2 Bootstrapping
• 3.6.3 Comparison of algorithms
• 3.6.3.1 K-Fold cv paired t-test
• 3.6.3.2 5 X 2 cv paired t-test
• 3.6.3.3 Combined 5 X 2 cv t-test
• 3.6.3.4 Combined 5 X 2 cv t-test
• References

Inspec keywords: learning (artificial intelligence)

Other keywords: information extraction problems; machine learning

Subjects: Knowledge engineering techniques