I enjoyed this post in Techemergence. I was originally looking for a definition of machine learning that helped reconcile all of the different definitions that are out there. I like the approach that they took. I modified a table they had (presented below) slightly to capture some of the thoughts on machine learning methods.
This is too complicated for many, but captures the idea that ML is layered, and will involve many techniques. I have a simplified list in Automating Data Management and Governance through Machine Learning.
See also No, Machine Learning is not just glorified Statistics for some more discussion in plain English on Machine learning.
| Classification | Scoring | Recommendation / Prediction |
| · K-Nearest Neighbor
· Support Vector Machines · Naïve Bayes · Logistic Regression · Decision Trees · Sets of Rules · Propositional Rules · Logic Rules · Neural Networks · Bayesian Networks · Conditional Random Fields |
· Accuracy / Error Rate
· Precision & Recall · Squared Error · Likelihood · Posterior Probability · Information Gain · K-L Divergence · Cost / Utility · Margin |
Combinatorial Optimization
· Greedy Search · Beam Search · Branch & Bound Continuous Optimization · Gradient Descent · Conjugant Gradient · Quasi Newton Method · Linear Programming · Non-Linear (Quadratic) Programming |
| Credit: Dr. Pedro Domingo, University of Washington (Slightly Simplified) | ||
analyticaltern 
One Response to “What is Machine Learning?”