Combining Pattern Classifiers: Methods and Algorithms, 2nd Edition

10 A FINAL THOUGHT

One of my friends who kindly read the manuscript and returned comments to me said: “Chapter 10. There isn’t one? No conclusion or retrospective on the past decade?”

Well, wise and prophetic words do not come easily to me. Only that famous Russian aphorism springs to mind which loosely means “you can’t embrace immensity,” or rather “you can’t embrace the unembraceable.” To me, the classifier ensemble field is starting to resemble the unembraceable. The lively interest in the area has led to significant advancements both in the structuring of the field and in populating this structure with exciting discoveries, ideas, methods, and results. However, it is impossible to compile a list of the most significant achievements in the past decade without offending or alienating half of my readership. I will leave this task to the main judge – time.

What is clear, though, is that classifier ensembles are here to stay. Solving the challenges of modern pattern recognition and machine learning, surfacing in application areas such as computer vision, medical research, geosciences, and many more, will require bespoke research arsenals. It is likely that the field will branch out, and the relevant fundamental research will be compartmentalized. I look forward to seeing this happen. I just hope we do not suffer the fate of those building the Tower of Babel, and we will still ...