A Deep Learning Approach for Molecular Classification Based on AFM Images
In spite of the unprecedented resolution provided by non-contact atomic force microscopy (AFM) with CO-functionalized and advances in the interpretation of the observed contrast, the unambiguous identification of molecular systems solely based on AFM images, without any prior information, remains an...
Main Authors: | , , |
---|---|
Format: | Article |
Language: | English |
Published: |
MDPI AG
2021-06-01
|
Series: | Nanomaterials |
Subjects: | |
Online Access: | https://www.mdpi.com/2079-4991/11/7/1658 |
id |
doaj-e771ac0261f74686a047eb6276d4f8ed |
---|---|
record_format |
Article |
spelling |
doaj-e771ac0261f74686a047eb6276d4f8ed2021-07-23T13:57:15ZengMDPI AGNanomaterials2079-49912021-06-01111658165810.3390/nano11071658A Deep Learning Approach for Molecular Classification Based on AFM ImagesJaime Carracedo-Cosme0Carlos Romero-Muñiz1Rubén Pérez2Quasar Science Resources S.L., Camino de las Ceudas 2, E-28232 Las Rozas de Madrid, SpainDepartment of Physical, Chemical and Natural Systems, Universidad Pablo de Olavide, Ctra. Utrera Km. 1, E-41013 Seville, SpainDepartamento de Física Teórica de la Materia Condensada, Universidad Autónoma de Madrid, E-28049 Madrid, SpainIn spite of the unprecedented resolution provided by non-contact atomic force microscopy (AFM) with CO-functionalized and advances in the interpretation of the observed contrast, the unambiguous identification of molecular systems solely based on AFM images, without any prior information, remains an open problem. This work presents a first step towards the automatic classification of AFM experimental images by a deep learning model trained essentially with a theoretically generated dataset. We analyze the limitations of two standard models for pattern recognition when applied to AFM image classification and develop a model with the optimal depth to provide accurate results and to retain the ability to generalize. We show that a variational autoencoder (VAE) provides a very efficient way to incorporate, from very few experimental images, characteristic features into the training set that assure a high accuracy in the classification of both theoretical and experimental images.https://www.mdpi.com/2079-4991/11/7/1658atomic force microscopy (AFM)deep learningmolecular recognitionvariational autoencoder (VAE) |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Jaime Carracedo-Cosme Carlos Romero-Muñiz Rubén Pérez |
spellingShingle |
Jaime Carracedo-Cosme Carlos Romero-Muñiz Rubén Pérez A Deep Learning Approach for Molecular Classification Based on AFM Images Nanomaterials atomic force microscopy (AFM) deep learning molecular recognition variational autoencoder (VAE) |
author_facet |
Jaime Carracedo-Cosme Carlos Romero-Muñiz Rubén Pérez |
author_sort |
Jaime Carracedo-Cosme |
title |
A Deep Learning Approach for Molecular Classification Based on AFM Images |
title_short |
A Deep Learning Approach for Molecular Classification Based on AFM Images |
title_full |
A Deep Learning Approach for Molecular Classification Based on AFM Images |
title_fullStr |
A Deep Learning Approach for Molecular Classification Based on AFM Images |
title_full_unstemmed |
A Deep Learning Approach for Molecular Classification Based on AFM Images |
title_sort |
deep learning approach for molecular classification based on afm images |
publisher |
MDPI AG |
series |
Nanomaterials |
issn |
2079-4991 |
publishDate |
2021-06-01 |
description |
In spite of the unprecedented resolution provided by non-contact atomic force microscopy (AFM) with CO-functionalized and advances in the interpretation of the observed contrast, the unambiguous identification of molecular systems solely based on AFM images, without any prior information, remains an open problem. This work presents a first step towards the automatic classification of AFM experimental images by a deep learning model trained essentially with a theoretically generated dataset. We analyze the limitations of two standard models for pattern recognition when applied to AFM image classification and develop a model with the optimal depth to provide accurate results and to retain the ability to generalize. We show that a variational autoencoder (VAE) provides a very efficient way to incorporate, from very few experimental images, characteristic features into the training set that assure a high accuracy in the classification of both theoretical and experimental images. |
topic |
atomic force microscopy (AFM) deep learning molecular recognition variational autoencoder (VAE) |
url |
https://www.mdpi.com/2079-4991/11/7/1658 |
work_keys_str_mv |
AT jaimecarracedocosme adeeplearningapproachformolecularclassificationbasedonafmimages AT carlosromeromuniz adeeplearningapproachformolecularclassificationbasedonafmimages AT rubenperez adeeplearningapproachformolecularclassificationbasedonafmimages AT jaimecarracedocosme deeplearningapproachformolecularclassificationbasedonafmimages AT carlosromeromuniz deeplearningapproachformolecularclassificationbasedonafmimages AT rubenperez deeplearningapproachformolecularclassificationbasedonafmimages |
_version_ |
1721286832702881792 |