You are currently browsing the tag archive for the ‘Hidden Markov model’ tag.
Computational biologists do not all recognize the Kalman filter by name, but they know it in the form of the hidden Markov model (the Kalman filter is a hidden Markov model with continuous latent variables and Gaussian observed variables). I mention this because while hidden Markov models, and more generally graphical models, have had an extraordinary impact on the tools and techniques of high-throughput biology, one of their primary conceptual sources, the Kalman filter, is rarely credited as such by computational biologists.
Illustration of the Kalman filter (from Wikipedia).
Where the Kalman filter has received high acclaim is in engineering, especially electrical and aeronautical engineering via its applications in control theory and where it has long been a mainstay of the fields. But it was not always so. The original papers, written in the early 1960s by Rudolf Kálmán and colleagues, were published in relatively obscure mechanical engineering journals rather than the mainstream electrical engineering journals of the time. This was because Kálmán’s ideas were initially scoffed at and rejected… literally. Kálmán second paper on the topic, New Results in Linear Filtering and Prediction Theory (with almost 6,000 citations), was rejected at first with a referee writing that “it cannot possibly be true”. The story is told in Grewal and Andrews’ book Kalman Filtering: Theory and Practice Using MATLAB. Of course not only was the Kalman filter theory correct, the underlying ideas were, in modern parlance, transformative and disruptive. In 2009 Rudolf Kálmán received the National Medal of Science from Barack Obama for his contribution. This is worth keeping in mind not only when receiving rejections for submitted papers, but also when writing reviews.
Rudolf Kálmán passed away at the age of 86 on Saturday July 2nd 2016.