Machine-Learning Techniques

TL;DR: These two broad classifications of machine-learning methods will ground us as the authors discuss a broad range of techniques and where they are currently being applied in life sciences research, expanding their toolkit and enabling us to take a very different path in their analysis efforts.

Abstract: Our ultimate objective in data mining is to identify any hidden patterns or relationships between our data elements, and in one sense, machine learning provides us with a set of techniques to do just that: techniques that allow us to learn the patterns without any outside influence (unsupervised learning). However, just as is the case with anything, that power comes at a price, but the results can be very interesting and very significant. In other cases, we have some sense on what the results should be and so can guide the learning techniques through an initial “training” phase, directing our system and honing the results (supervised learning). These two broad classifications of machine-learning methods will ground us as we discuss a broad range of techniques and where they are currently being applied in life sciences research, expanding our toolkit and enabling us to take a very different path in our analysis efforts: using an artificial intelligence discipline and letting the data tell us what it contains. As datasets grow, these techniques become more important.