Having recently attended a Biometrics conference where one of Vivent’s collaborators was presenting results on the analysis of plant electrophysiology signals captured using PhytlSigns sensors, I was surprised at the amount of time and energy invested in discussing the differences between machine learning methods and statistics. It didn’t take long to realise that I’d just stumbled into a long and contentious debate that’s been running for years in data science circles.
Using machine learning techniques has enabled Vivent to make some important breakthroughs in understanding plant electrophysiology signals and have enabled us to direct our application of statistical methods more effectively. For me, the approaches have always been complementary so the vigorous debate was a little surprising.
If you are interested in understanding the similarities with and differences between statistics and machine learning then I recommend this blog post which clearly explains distinctions in objectives, approaches and applications. The post references a Nature article which is also clearly written.
The Actual Difference Between Statistics and Machine Learning by Matthew Stewart
Whether I am writing a job description or an academic paper I will now be much more precise with the language I am using to describe our signal processing and analysis.