IAES International Journal of Artificial Intelligence (IJ-AI)

Vol. 12, No. 4, December 2023, pp. 2042~2047

ISSN: 2252-8938, DOI: 10.11591/ijai.v12.i4.pp2042-2047  2042

Modelling mechanisms for measurable and detection based on

artificial intelligence

Raghad AbdulHadi AbdulQader, Marwa J. M. Zedan

College of Computer Sciences and Mathematics, Department of Computer Science, Mosul University, Iraq

Article Info ABSTRACT

Article history: One of the trendiest areas in the field of materials science is Artificial
Intelligence (AI) based physical applications. Typically, more time and
Received Nov 8, 2022 resources are needed for traditional experiments and statistical methods. Thus,
Revised Feb 27, 2023 there is a growing need for applications of AI in the simulation and
Accepted Mar 10, 2023 investigation of novel materials. Usually, there are significant restrictions
because there are not any benchmark datasets, sophisticated pre-processing
mechanisms, prediction modelling mechanisms, or simulation tools in the
Keywords: literature on materials. This work aims to attempt for examining
computational and experimental data-based AI processes. In addition, the
AI scenarios state of research into developing new materials and utilizing AI in material
Analysis of materials modelling tools is implemented. As long as, AI can be used in materials to
Optimization models improve efficiency and prediction accuracy. Also, it is very difficult to
determine great learning models, involving data preparation, model
architecture, data management, and simulation techniques. Finally, it has been
discussed the challenges in realizing AI-based applications in the field of
materials science.
This is an open access article under the CC BY-SA license.

Corresponding Author:
Raghad AbdulHadi AbdulQader
College of Computer Sciences and Mathematics, Department of Computer Science, Mosul University
Mosul, Iraq
Email: [email protected]

1. INTRODUCTION
Science is generally a gradual process; a product that is ready for sale can take months or even years
to develop from the investigation on, say, a novel material notion. Artificial intelligence (AI) advancements,
however, have the potential to significantly speed up that laborious process. The path of experiment, modelling,
and simulation, as well as exploration and comprehension, are all aided by computer algorithms. They are
searching for and perfecting new materials for the technologies of the future in tandem with human intelligence
and creativity [1].
The materials genome is one initiative that has been launched to utilize computational and data-driven
resources [2]. In addition, applications of AI are fundamentally altering every aspect of life, particularly in the
technological sphere [3]. The demand for AI in the modelling and investigation of novel ceramic materials is
rising as a result of its effective applicability for creating efficiency and performance. It is anticipated that
materials development based on AI analysis will produce novel materials and lower the development cost both
in terms of time and materials [4], [5]. The scientific community has, however, noted numerous constraints on
the discovery and use of improved materials based on advanced machine learning and AI approaches. For
instance, computational simulation has several challenges, and high-performance index features are needed for
the materials' structures. To generate a fundamental understanding of the input variable conditions and

Journal homepage: http://ijai.iaescore.com

Int J Artif Intell ISSN: 2252-8938  2043

efficiency index qualities, innovative materials research built on the convergence of AI approaches and
empirical mechanisms is required [6].
In other words, AI and machine learning working together have sparked a huge change in several
fields, including the detection of different materials. In the beginning, symbolic techniques were chosen to
uncover hidden knowledge in the data. Later, the methods were adjusted to include some cutting-edge features
in artificial neural networks, which can train on their own with the aid of neurons. Similar to this, various other
practical machine learning models, such as support vector machines (SVMs), and decision trees (DTs), were
also introduced. Industry and academics are currently interested in a variety of revolutionary machine learning-
based techniques like deep learning to evaluate vast amounts of data [7]. As a result, the machine learning
paradigm aids in the automation of the activity of analytical building construction. With the advent of machine
learning algorithms that extract from the data iteratively, the difficulty of explicit programming to expose the
hidden patterns in the information has been reduced. High-dimensional data can be handled by a variety of
machine learning techniques, such as clustering, modelling, and categorization. Machine learning-based
models' main objective is to thoroughly examine enormous databases to uncover hidden knowledge. In a
variety of fields, including machine vision, natural language processing (NLP), and data security, machine
learning (ML) algorithms adapt from the previous data to produce accurate and dependable results. Machine
learning-based models are used to solve a variety of routine tasks, including sentiment analysis, fraud detection,
and web searches [8].
Figure 1, shows understanding the forces that regulate phenomena is essential for the advancement of
materials research since it enables the discovery and application optimization of new materials. Fundamentally,
data from experiments and simulations for substances synthesized under various conditions must be made
available to access the knowledge space and speed up this cycle. Imaging offers a window into specific contexts
and creates a vital connection for comprehending the causes that underlie observed behaviour. These databases
may be created thanks to ML technologies, which also make it easier to forecast attributes from data-driven
models quickly. Similar to this, the data can be combined using statistical mechanical models or a Bayesian
formulation to aggregate all accessible sources of information and generate more accurate predictions. The
acquired knowledge should be transferrable and allow for more effective design cycles for related material
systems. These technologies all require community initiatives for the accessibility of software, data, and
procedures, which is vital to creating this new future.

Figure 1. Shows the general AI mechanism for simulation and modeling [4]

Modelling mechanisms for measurable and detection based on… (Raghad AbdulHadi AbdulQader)
2044  ISSN: 2252-8938

2. MODELLING MECHANISMS
Over the last few years, interdisciplinary research and applications have grown more interested in
modelling mechanisms [9]. Material applications based on artificial intelligence processes have lately surfaced
with increasing computational and experimental data due to interdisciplinary study interest. Utilizing current
material data to anticipate the properties of novel materials using data science techniques and mathematics is
a key task of substance science-based AI technologies [10]. Building a classifier model that can forecast the
desired attribute from a known collection of input substance features is the first stage. For instance, one of the
important descriptor models with input parameters that represent material structure features is the quantitative
structure-property link (QSPR). It is difficult for conventional linear and non-linear operations to handle a
complicated relationship between the input and result of material properties. However, ML techniques can now
effectively represent these complex interactions [11]. Traditional linear and non-linear correlation approaches
have difficulty handling a complicated connection between the input and outputs of material attributes.
However, ML techniques may now primary products of these complex relationships [12]. A model is employed
to anticipate material behaviour through a material process model based on AI or statistics after doing adjective
modelling analysis and becoming familiar with the data. Intelligent machine learning (ML) models use training
data to discover patterns in data to automatically enhance performance. The various machine learning
algorithms are classified below by kind. The predictive technique applied in applications for material science
is shown in Table 1.

Table 1. Applications of predictive mechanisms in material science [5]

No. Approaches Types
1. “Least-squares regression” Deterioration
2. “Kernel ridge regression” Deterioration
3. “Kriging or Gaussian process regression” Deterioration
4. “Artificial Neural Network” Deterioration Arrangement
5. “Sustenance Vector Machine” Deterioration Arrangement
6. “Result tree” Arrangement
7. “Random forest” Arrangement
8. “k-nearest neighbours” Arrangement
9. Simple Bayes Arrangement

For example, Naive Bayes classification ML implementations in material research include

bandgaps [13], alloys [14], and steel strength [15] property prediction. Catalytic activity [16] and equilibrium
constant constants [17], [18] are used in the selection of stable materials. Polymer dielectrics and mixed oxide
reactors are two applications of polymer-based materials research [19], [20].
Unsupervised learning uncovers the connections between the data, while supervised learning
identifies a mapping function input data to an output property. In unsupervised learning, clustering separates a
dataset into groups so that the data points within a bunch are extra compared to one another than those inside
other clusters. The best method for deriving scientific insights from data and discovering novel, interesting
substances based on relative studies is clustering. General clustering procedures include K-means, clustering,
and hidden Markov modelling [21], [22]. The science world also examines recent uses of various ML
techniques and ML technologies for material discovery. A few published evaluations on AI for materials,
however, have focused on a single composite structure or certain techniques. As a result, the research focuses
on an application-based approach to material discovery that is boosted by AI. Property prediction,
characterisation, synthesis, and theoretical paradigm discovery are some of the AI techniques discussed in the
study. Table 2, shows a list of datasets organized by titles, physical characteristics, size, and AI jobs. Based on
titles, material characteristics, dataset size, and AI tasks. The categorical dataset populations of the dataset were
gathered using techniques for ML and AI that have been described in the literature on materials science. To
the left of the classification techniques employed in the graph, there is a list of descriptors over each colour
bar, and each bar indicates the number of data. The large and small data sets, which each contain 100 and 1000
occurrences, respectively, range in size from 100 to 5000. A significant portion of the data taken from this data
falls under the category of computed data, which includes experimental data and calculated data [16]–[23].
The three features of the mentioned data sets are explained: The term "Material Property" describes
the characteristics of the material, such as the "Lonic Dielec Const." Additionally, the name of the primary
research publication for each dataset is cited. Additionally, the name of each dataset's underlying research
publication is cited.

Int J Artif Intell, Vol. 12, No. 4, December 2023: 2042-2047

Int J Artif Intell ISSN: 2252-8938  2045

Table 2. A list of datasets organised by name, composition, size, and AI tasks

No. Material Property Dataset Size All Task Type
1. Bandgap (Eg) 1266 Deterioration
2. Affinity Formation Energy Lattice 171 Deterioration
3. Spring Constant 166 Deterioration
4. Total Dielec Const 169 Deterioration
5. Interfacial Energy Lattice Parameter Formation Energy” 1241 Deterioration
6. GGA Bandgap 1229 Deterioration
7. HSE Bandgap 111 Deterioration
8. Elastic Constants: c11, c12, c13, c33, c44 990 Deterioration
9. Effective Thermal Conductivity 711 Deterioration
10. Effective Thermal Conductivity 323 Deterioration
11. Elastic Moduli: Shear Modulus (G), Bulk Modulus (K) 5468 Deterioration
12. Curie Temperature (Tc) 189 Classification
13. Curie Temperature (Tc) 124 Deterioration
14. Bandgap (G, Wo)” 255 Deterioration
15. Ehull 1920 Classification
16. Electric Dielec. Const, Bandgap 263 Deterioration
17. Lonic Dielec. Const 151 Deterioration
18. Melting Temperature (Tm) 250 Deterioration
19. Bandgap Electric Dielec. Const 6254 Deterioration
20. Glass Transition Temp (Tg) 26 Classification

3. SIMULATION TOOLS AND RESULTS

In the literature, a lot of machines having to learn simulation tools have been developed. In this part,
we provide a quick overview of simulation tools that have been published in reputable magazines like Nature.
Comparing traditional ML techniques for predicting properties of materials from elemental analysis with a
detailed understanding of material chemistry from simply its elemental composition [24], [25]. Through the
creation and application of deep neural networks, ElemNet is employed to dynamically capture the similarities
and interactions between physical and chemical processes. ElemNet allows for quick and reliable screening of
potential novel material candidates across a wide combinatorial space. Comparison of a deep learning method
with a traditional ML strategy for materials property prediction ElemNet predicts huge numbers of chemical
systems that could contain unidentified substances [26]. Deep learning allows for the avoidance of manual
feature building that requires domain expertise, producing superior outcomes even with a small number of
training examples. To facilitate data-driven studies and material property predictions, the open-source software
platform matminer was created. Large data sets can be imported from outside data sources using Matminer's
python modules. Citrination, materials research, materials gateway for data scientists, and materials data
factory databases serve as the foundation for these data sources. Additionally, it offers an application
programming interface (API) for putting code from a sizable feature extraction library created by the materials
science world into practice [27]. Some of the simulation tools for analysing material properties and structures
are listed in Table 3 for public use.
This part includes a thorough discussion as well as an explanation of the research's findings. Figure 1
and tables, and other easy-to-understand formats can be used to show results [6], [7]. There are various ways
to break up the topic.

Table 3. Tools for materials analysis simulation are obtainable to the community
No. Name Explanation
1. AFLOW Online applications for property predictions using machine learning
2. Artificial Neural Network Regression Classification
3. Support Vector Machine Regression Classification
4. k-nearest neighbours Classification
5. CALPHAD Computer Coupling of Phase Diagrams and Thermochemistry
6. Matminer Data source, descriptive and predictive analysis
7. Decision tree Classification
8. ElemNet Deep learning-based appliance
9. ChemSpider Exploration machine for Interaction's construction database
10. Citrination AI-Powered Resources Data Stage

4. CONCLUSION
Materials scientists and members of newly formed interdisciplinary communities have paid particular
attention to AI-based techniques. AI and its subfields, including machine learning, can be used to examine big
data derived from actual databases and datasets. These tools provide associations between a wide range of
complexes and associated structural components in the composition of materials. This review paper compiles
Modelling mechanisms for measurable and detection based on… (Raghad AbdulHadi AbdulQader)
2046  ISSN: 2252-8938

the implementations of material science employing AI-based modelling techniques and materials modelling
tools. To predict changes in certain parameters for modelling the behaviour of materials, such as their mixes,
situations, and functional properties, Advanced technologies like deep learning show significant breakthroughs
and potential. The analysis of AI approaches shows the value of sophisticated AI mechanism-based models for
developing and improving characteristics prediction for the development of novel materials. In our forthcoming
work, we'll offer a framework for publishing and analyzing content based on composition assessment and deep
learning for efficient content discovery.

ACKNOWLEDGEMENTS
Great thanks to the University of Mosul for its support and assistance during making this research.

REFERENCES
[1] P. Ball, “Using artificial intelligence to accelerate materials development,” MRS Bulletin, vol. 44, no. 5, pp. 335–343, 2019,
doi: 10.1557/mrs.2019.113.
[2] Imran, N. Iqbal, S. Ahmad, and D. H. Kim, “Towards mountain fire safety using fire spread predictive analytics and mountain fire
containment in iot environment,” Sustainability (Switzerland), vol. 13, no. 5, pp. 1–23, 2021, doi: 10.3390/su13052461.
[3] M. Eminagaoglu and S. Eren, “Implementation and comparison of machine learning classifiers for information security risk analysis
of a human resources department,” in 2010 International Conference on Computer Information Systems and Industrial Management
Applications, CISIM 2010, 2010, vol. 3, pp. 187–192. doi: 10.1109/CISIM.2010.5643665.
[4] R. K. Vasudevan et al., “Materials science in the artificial intelligence age: High-throughput library generation, machine learning,
and a pathway from correlations to the underpinning physics,” MRS Communications, vol. 9, no. 3, pp. 821–838, 2019, doi:
10.1557/mrc.2019.95.
[5] Imran and D. H. Kim, “Artificial intelligence-based modeling mechanisms for material analysis and discovery,” Journal of
Intelligent Pervasive and Soft Computing, vol. 1, no. 01, pp. 10–15, 2022, [Online]. Available: Artificial Intelligence-based
Modeling Mechanisms for Material %0AAnalysis and Discovery
[6] H. Margossian, G. Deconinck, and J. Sachau, “Distribution network protection considering grid code requirements for distributed
generation,” IET Generation, Transmission and Distribution, vol. 9, no. 12, pp. 1377–1381, Sep. 2015, doi: 10.1049/iet-
gtd.2014.0987.
[7] O. Núñez-Mata, R. Palma-Behnke, F. Valencia, A. Urrutia-Molina, P. Mendoza-Araya, and G. Jiménez-Estévez, “Coupling an
adaptive protection system with an energy management system for microgrids,” Electricity Journal, vol. 32, no. 10, p. 106675, Dec.
2019, doi: 10.1016/j.tej.2019.106675.
[8] Imran, N. Iqbal, and D. H. Kim, “IoT task management mechanism based on predictive optimization for efficient energy
consumption in smart residential buildings,” Energy and Buildings, vol. 257, p. 111762, Feb. 2022,
doi: 10.1016/j.enbuild.2021.111762.
[9] Imran, F. Jamil, and D. Kim, “An ensemble of a prediction and learning mechanism for improving accuracy of anomaly detection
in network intrusion environments,” Sustainability (Switzerland), vol. 13, no. 18, p. 10057, Sep. 2021, doi: 10.3390/su131810057.
[10] U. Zaman, Imran, F. Mehmood, N. Iqbal, J. Kim, and M. Ibrahim, “Towards secure and intelligent internet of health things: a survey
of enabling technologies and applications,” Electronics (Switzerland), vol. 11, no. 12, p. 1893, Jun. 2022,
doi: 10.3390/electronics11121893.
[11] Imran, F. Qayyum, D. H. Kim, S. J. Bong, S. Y. Chi, and Y. H. Choi, “A survey of datasets, preprocessing, modeling mechanisms,
and simulation tools based on ai for material analysis and discovery,” Materials, vol. 15, no. 4, p. 1428, Feb. 2022,
doi: 10.3390/ma15041428.
[12] “(PDF) An artificial intelligence approach in designing new materials.”
https://www.researchgate.net/publication/42253244_An_artificial_intelligence_approach_in_designing_new_materials (accessed
Jul. 03, 2023).
[13] R. Ramprasad, R. Batra, G. Pilania, A. Mannodi-Kanakkithodi, and C. Kim, “Machine learning in materials informatics: Recent
applications and prospects,” npj Computational Materials, vol. 3, no. 1, Dec. 2017, doi: 10.1038/s41524-017-0056-5.
[14] N. Iqbal et al., “Enhanced time-constraint aware tasks scheduling mechanism based on predictive optimization for efficient load
balancing in smart manufacturing,” Journal of Manufacturing Systems, vol. 64, pp. 19–39, Jul. 2022,
doi: 10.1016/j.jmsy.2022.05.015.
[15] A. M. Virshup, J. Contreras-García, P. Wipf, W. Yang, and D. N. Beratan, “Stochastic voyages into uncharted chemical space
produce a representative library of all possible drug-like compounds,” Journal of the American Chemical Society, vol. 135, no. 19,
pp. 7296–7303, May 2013, doi: 10.1021/ja401184g.
[16] T. Zhou, Z. Song, and K. Sundmacher, “Big data creates new opportunities for materials research: a review on methods and
applications of machine learning for materials design,” Engineering, vol. 5, no. 6, pp. 1017–1026, Dec. 2019,
doi: 10.1016/j.eng.2019.02.011.
[17] A. Jain et al., “Commentary: The materials project: A materials genome approach to accelerating materials innovation,” APL
Materials, vol. 1, no. 1, Jul. 2013, doi: 10.1063/1.4812323.
[18] M. T. Michalski RS, Carbonell JG, ML: an AI approach. Berlin: Springer-Verlag, 2013.
[19] A. Agrawal, P. D. Deshpande, A. Cecen, G. P. Basavarsu, A. N. Choudhary, and S. R. Kalidindi, “Exploration of data science
techniques to predict fatigue strength of steel from composition and processing parameters,” Integrating Materials and
Manufacturing Innovation, vol. 3, no. 1, pp. 90–108, Apr. 2014, doi: 10.1186/2193-9772-3-8.
[20] S. K. Karak, S. Chatterjee, and S. Bandopadhyay, “Mathematical modelling of the physical and mechanical properties of nano-
Y2O3 dispersed ferritic alloys using evolutionary algorithm-based neural network,” Powder Technology, vol. 274, pp. 217–226,
Apr. 2015, doi: 10.1016/j.powtec.2015.01.028.
[21] G. Pilania, A. Mannodi-Kanakkithodi, B. P. Uberuaga, R. Ramprasad, J. E. Gubernatis, and T. Lookman, “Machine learning
bandgaps of double perovskites,” Scientific Reports, vol. 6, no. 1, Jan. 2016, doi: 10.1038/srep19375.

Int J Artif Intell, Vol. 12, No. 4, December 2023: 2042-2047

Int J Artif Intell ISSN: 2252-8938  2047

[22] R. Jinnouchi and R. Asahi, “Predicting Catalytic Activity of Nanoparticles by a DFT-Aided Machine-Learning Algorithm,” Journal
of Physical Chemistry Letters, vol. 8, no. 17, pp. 4279–4283, Aug. 2017, doi: 10.1021/acs.jpclett.7b02010.
[23] T. Zhou, S. Jhamb, X. Liang, K. Sundmacher, and R. Gani, “Prediction of acid dissociation constants of organic compounds using
group contribution methods,” Chemical Engineering Science, vol. 183, pp. 95–105, Jun. 2018, doi: 10.1016/j.ces.2018.03.005.
[24] M. Z. Aghaji, M. Fernandez, P. G. Boyd, T. D. Daff, and T. K. Woo, “Quantitative structure–property relationship models for
recognizing metal organic frameworks (MOFs) with high CO2Working capacity and CO2/CH4Selectivity for methane
purification,” European Journal of Inorganic Chemistry, vol. 2016, no. 27, pp. 4505–4511, Jun. 2016, doi: 10.1002/ejic.201600365.
[25] V. Sharma et al., “Rational design of all organic polymer dielectrics,” Nature Communications, vol. 5, no. 1, Sep. 2014,
doi: 10.1038/ncomms5845.
[26] D. Jha et al., “ElemNet: deep learning the chemistry of materials from only elemental composition,” Scientific Reports, vol. 8,
no. 1, Dec. 2018, doi: 10.1038/s41598-018-35934-y.
[27] L. Ward et al., “Matminer: An open source toolkit for materials data mining,” Computational Materials Science, vol. 152,
pp. 60–69, Sep. 2018, doi: 10.1016/j.commatsci.2018.05.018.

BIOGRAPHIES OF AUTHORS

Raghad Abdul Hadi Abdul Qader She has been assistant literature at the
department of computer sciences, college of computer science and mathematics, the University
of Mosul, Iraq since 2021, graduated from the computer science and mathematics college at the
University of Mosul, Iraq in 2000, and worked as a programmer in the same Collage until 2019
when she also started studying Masters of science in same collage, then she finished MSC.
Degree in 2021. My general expertise is computer science, and my speciality is in the area of
artificial intelligence and cryptography. She has a research gate account under the name Raghad
Abdul Hadi. She can be contacted at email: [email protected].

Marwa Jassim Mohammad She has been assistant literature at the department of
computer sciences, college of computer science and mathematics, the University of Mosul, Iraq
since 2020, graduated from the computer science and mathematics college at the University of
Mosul, Iraq in 2001, and worked as a programmer in the same Collage until 2018 when she
also started studying Masters of science in same collage, then she finished MSC. Degree in
2020, my general expertise in computer science, and my speciality are in the area of artificial
intelligence and Image Processing. She has a research gate account under the name Marwa J.M.
Zedan. She can be contacted at email: [email protected].

Modelling mechanisms for measurable and detection based on… (Raghad AbdulHadi AbdulQader)