Bayesian updating and and Bayesian networks are used in so many machine learning algorithms, but can we consider them as machine learning techniques? and if we can, under what circumstances? and in what kind of applications?
The root of Bayesian inference are in statistics, even if nowadays it is legitimately considered part of the broader field of data science, since the border b/w statistics and machine learning/statistical learning is blurred. If you want to draw a line, we can say that Naive Bayes is the typical approach of machine learning, rather than classic statistics, since in the former less assumptions (eg on the independence of predictors) are made, while in the latter assumptions are stronger (eg variables must be independent etc.). Still, the naive approach gives good results empirically , given that data are large.
I agree with Davide Barbieri s points. I would argue that Bayesian methods are becoming more and more used due to the available computational power.
Davide Barbieri's point on this matter is quite clear and revealing. Additionally, the notion in Bayesian statistics needs to be looked at separately from the frequentists' approach. The basic ideology here is that we look at the variability in distribution rather than the data set. Therefore, IMO the Bayesian methods are closer to machine learning "technique". Yet they are not the same. Machine learning should be considered as a statistical tool whereas Bayesian methods should fall under mathematical models.
- Badges
- Science topic
- Similar topics
- Medicine
- Physiology
- Biosignals
- Biomedical Signal Processing
- Statistical Learning