Simple machine learning methods work surprisingly well for Ramanomics

Aug 4, 2026

Ramanomics studies have reported high performance using a deep convolutional neural network for liver cancer detection, yet the author’s experience on another Ramanomics dataset suggests simpler models can match or exceed deep learning. Logistic regression achieved about 90.4% accuracy in under a minute, while a random decision forest reached about 92.6% accuracy in under 10 seconds. The work argues deep learning remains promising but has not delivered a major performance leap. More biophysics-aware machine learning is advocated.

LanguageEnglish

CategoryTechnology

FormatPDF

Pages5

Reading Time13 min

AccessFree to download

Short communication

Abstract

Introduction

Results

University of Groningen Simple machine learning methods work surprisingly well for Ramanomics Lawrence, Celestine P. Published in: Journal of Raman Spectroscopy DOI: 10.1002/jrs.6555 IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below. Document Version Publisher's PDF, also known as Version of record Publication date: 2023 Link to publication in University of Groningen/UMCG research database Citation for published version (APA): Lawrence, C. P. (2023) . Simple machine learning methods work surprisingly well for Ramanomics. Journal of Raman Spectroscopy, 54(8), 887-889 . [https://doi.org/10.1002/jrs.6555](https://doi.org/10.1002/jrs.6555) Copyright Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons) . The publication may also be distributed here under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license. More information can be found on the University of Groningen website: [https://www.rug.nl/library/open-access/self-archiving-pure/taverne](https://www.rug.nl/library/open-access/self-archiving-pure/taverne)amendment. Take-down policy If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim. Downloaded from the University of Groningen/U MCG research database (Pure): [http://www.rug. nl/research/portal. For technical reasons the](http://www.rug. nl/research/portal. For technical reasons the)[ ](http://www.rug. nl/research/portal. For technical reasons the)[number of authors shown on this cover page is limited to 10 maximum.](number of authors shown on this cover page is limited to 10 maximum.) Download date: 03-08-2026 Received: 15 February 2023 Revised: 4 May 2023 Accepted: 5 May 2023 DOI: 10.1002/jrs.6555 SHO RT C OMM UNICA TI ON Simple machine learning methods work surprisingly well for Ramanomics Celestine P. Lawrence Bernoulli Institute and Groningen Cognitive Systems and Materials Center (CogniGron), University of Groningen, Groningen, Netherlands Correspondence Celestine P. Lawrence, Bernoulli Institute and Groningen Cognitive Systems and Materials Center (CogniGron), University of Groningen, 9700 AB, Groningen, Netherlands. Email: [c.p.lawrence@rug.nl](c.p.lawrence@rug.nl) Abstract Recently, a deep convolutional neural network was employed to detect liver cancer by Ramanomics. Results on a demo dataset claimed an accuracy of about 90% to be achieved after around an hour of training in a modern desktop. However, my experience with another Ramanomics dataset taught me that simple methods could potentially outperform deep learning. Here, I tested the simple and interpretable method of logistic regression. It achieved an accuracy of around 90.4% in under a minute. Employing a random decision forest yieldsan accuracy of 92.6% in under 10 s. Thus, although deep learning is promising, it is yet to provide a quantum-leap in performance for Ramanomics. A biophysics aware machine learning method would be more welcome! KEYWOR DS machine learning, Raman spectroscopy 1 | INTRODUCTION Classification of Raman spectra by deep learning methods has been a topic of interest for over three decades.1 With the recent success of deep learning, in scientific domains such as protein folding,2 there also seems to be a growing wave of interest in deep learning for Ramanomics.3,4 However, simpler machine learning methods could potentially outperform deep learning methods.5 I personally found this to be true, on a dataset in the wild, when I could gain a 0.5% in accuracy for the classification of pathogenic bacteria by their Raman spectr

FAQs

What models were tested in the Ramanomics classification task?

The author tested logistic regression and a random decision forest to classify Raman spectra.

How does the performance of simple methods compare with deep learning?

Logistic regression achieved around 90.4% accuracy in under a minute, and the random decision forest reached around 92.6% accuracy in under 10 seconds, outperforming the previously proposed deep neural network on the referenced task.

What conclusion is drawn about future Ramanomics machine learning?

Deep learning is considered promising, but the author argues it has not yet provided a quantum-leap in performance; a more biophysics-aware approach is preferred.

Aug 4, 2026

Upload by Lucas Martin