Neural networks: A new method for solving chemical problems or just a passing phase?

AI in computational chemistry through the lens of a decade-long journey

This article gives a perspective on the progress of AI tools in computational chemistry through the lens of the author's decade-long contributions put in the wider context of the trends in this rapidly expanding field. This progress over the last decade is tremendous: while a decade ago we had a glimpse of what was to come through many proof-of-concept studies, now we witness the emergence of many AI-based computational chemistry tools that are mature enough to make faster and more accurate simulations increasingly routine. Such simulations in turn allow us to validate and even revise experimental results, deepen our understanding of the physicochemical processes in nature, and design better materials, devices, and drugs. The rapid introduction of powerful AI tools gives rise to unique challenges and opportunities that are discussed in this article too.

Artificial Intelligence and Advanced Materials

Artificial intelligence (AI) is gaining strength, and materials science can both contribute to and profit from it. In a simultaneous progress race, new materials, systems, and processes can be devised and optimized thanks to machine learning (ML) techniques, and such progress can be turned into innovative computing platforms. Future materials scientists will profit from understanding how ML can boost the conception of advanced materials. This review covers aspects of computation from the fundamentals to directions taken and repercussions produced by computation to account for the origins, procedures, and applications of AI. ML and its methods are reviewed to provide basic knowledge of its implementation and its potential. The materials and systems used to implement AI with electric charges are finding serious competition from other information-carrying and processing agents. The impact these techniques have on the inception of new advanced materials is so deep that a new paradigm is developing where implicit knowledge is being mined to conceive materials and systems for functions instead of finding applications to found materials. How far this trend can be carried is hard to fathom, as exemplified by the power to discover unheard of materials or physical laws buried in data.

The Evolution of Data-Driven Modeling in Organic Chemistry

Organic chemistry is replete with complex relationships: for example, how a reactant's structure relates to the resulting product formed; how reaction conditions relate to yield; how a catalyst's structure relates to enantioselectivity. Questions like these are at the foundation of understanding reactivity and developing novel and improved reactions. An approach to probing these questions that is both longstanding and contemporary is data-driven modeling. Here, we provide a synopsis of the history of data-driven modeling in organic chemistry and the terms used to describe these endeavors. We include a timeline of the steps that led to its current state. The case studies included highlight how, as a community, we have advanced physical organic chemistry tools with the aid of computers and data to augment the intuition of expert chemists and to facilitate the prediction of structure-activity and structure-property relationships.

Machine Learning Meets with Metal Organic Frameworks for Gas Storage and Separation

The acceleration in design of new metal organic frameworks (MOFs) has led scientists to focus on high-throughput computational screening (HTCS) methods to quickly assess the promises of these fascinating materials in various applications. HTCS studies provide a massive amount of structural property and performance data for MOFs, which need to be further analyzed. Recent implementation of machine learning (ML), which is another growing field in research, to HTCS of MOFs has been very fruitful not only for revealing the hidden structure-performance relationships of materials but also for understanding their performance trends in different applications, specifically for gas storage and separation. In this review, we highlight the current state of the art in ML-assisted computational screening of MOFs for gas storage and separation and address both the opportunities and challenges that are emerging in this new field by emphasizing how merging of ML and MOF simulations can be useful.

Molecular excited states through a machine learning lens

Theoretical simulations of electronic excitations and associated processes in molecules are indispensable for fundamental research and technological innovations. However, such simulations are notoriously challenging to perform with quantum mechanical methods. Advances in machine learning open many new avenues for assisting molecular excited-state simulations. In this Review, we track such progress, assess the current state of the art and highlight the critical issues to solve in the future. We overview a broad range of machine learning applications in excited-state research, which include the prediction of molecular properties, improvements of quantum mechanical methods for the calculations of excited-state properties and the search for new materials. Machine learning approaches can help us understand hidden factors that influence photo-processes, leading to a better control of such processes and new rules for the design of materials for optoelectronic applications.

Machine Learning in Computational Surface Science and Catalysis: Case Studies on Water and Metal-Oxide Interfaces

The goal of many computational physicists and chemists is the ability to bridge the gap between atomistic length scales of about a few multiples of an Ångström (Å), i. e., 10⁻¹⁰ m, and meso- or macroscopic length scales by virtue of simulations. The same applies to timescales. Machine learning techniques appear to bring this goal into reach. This work applies the recently published on-the-fly machine-learned force field techniques using a variant of the Gaussian approximation potentials combined with Bayesian regression and molecular dynamics as efficiently implemented in the Vienna ab initio simulation package, VASP. The generation of these force fields follows active-learning schemes. We apply these force fields to simple oxides such as MgO and more complex reducible oxides such as iron oxide, examine their generalizability, and further increase complexity by studying water adsorption on these metal oxide surfaces. We successfully examined surface properties of pristine and reconstructed MgO and Fe₃O₄ surfaces. However, the accurate description of water–oxide interfaces by machine-learned force fields, especially for iron oxides, remains a field offering plenty of research opportunities.

The analysis of similarities between the European Union countries in terms of the level and structure of the emissions of selected gases and air pollutants into the atmosphere

Based on the newly adopted strategy "The European Green Deal", by 2050, the European Union should become the first climate neutral region worldwide. This very ambitious goal will require many political, social and economic activities. Huge financial resources will also be needed to change the economy in order to reduce the emissions of harmful substances into the environment. The implementation of such an ambitious climate policy requires the development of a very reasonable economic plan, backed by many analyses, to ensure adequate financing of this idea. One of the basic objectives of such a plan should be to appropriately target aid funds to a group of countries with a similar structure of the emissions in question. The identification of the groups of similar countries in terms of the structure of harmful substance emissions requires the development of both appropriate methodology and applicable studies. Such methodology is presented in this paper, namely the Kohonen's artificial neural network model. The main objective of the developed methodology was to divide the European Union countries into groups similar in terms of the emissions of selected gases and dusts into the atmosphere. In addition to the division of the European Union countries into similar groups by the total volume of the emissions of studied substances, completely new division criteria were introduced. It was assumed that in order for the results of this study to be practically used, it is necessary to broaden the scope of the analysis. Therefore, an additional division of the European Union countries was made in relation to the volume of the emissions per capita, the value of gross domestic product and the area of a given country. This new approach was intended to show the diversity of the European Union countries in economic, demographic and geographical terms. The grouping results should be regarded as additional information to be utilized when preparing specific action plans to improve the state of the environment. Definitely, these plans need to be dedicated both to the groups of countries and the entire sectors in these groups. This will enable the efficient use of financial resources and can be a huge impetus for the European Union economic development. It will also allow smaller and less prosperous countries to achieve their goals. Undoubtedly, the developed methodology and conducted research allowed the authors to solve a significant research problem, and the results can be successfully used in practice.

Neural Network Potential Energy Surfaces for Small Molecules and Reactions

We review progress in neural network (NN)-based methods for the construction of interatomic potentials from discrete samples (such as ab initio energies) for applications in classical and quantum dynamics including reaction dynamics and computational spectroscopy. The main focus is on methods for building molecular potential energy surfaces (PES) in internal coordinates that explicitly include all many-body contributions, even though some of the methods we review limit the degree of coupling, due either to a desire to limit computational cost or to limited data. Explicit and direct treatment of all many-body contributions is only practical for sufficiently small molecules, which are therefore our primary focus. This includes small molecules on surfaces. We consider direct, single NN PES fitting as well as more complex methods that impose structure (such as a multibody representation) on the PES function, either through the architecture of one NN or by using multiple NNs. We show how NNs are effective in building representations with low-dimensional functions including dimensionality reduction. We consider NN-based approaches to build PESs in the sums-of-product form important for quantum dynamics, ways to treat symmetry, and issues related to sampling data distributions and the relation between PES errors and errors in observables. We highlight combinations of NNs with other ideas such as permutationally invariant polynomials or sums of environment-dependent atomic contributions, which have recently emerged as powerful tools for building highly accurate PESs for relatively large molecular and reactive systems.

Gas sensors based on mass-sensitive transducers. Part 2: Improving the sensors towards practical application

Within the framework outlined in the first part of the review, the second part addresses attempts to increase receptor material performance through the use of sensor systems and chemometric methods, in conjunction with receptor preparation methods and sensor-specific tasks. Conclusions are then drawn, and development perspectives for gravimetric sensors are discussed.

A Bayesian Regularized Artificial Neural Network for Simultaneous Determination of Loratadine, Naproxen and Diclofenac in Wastewaters

Background::

Simultaneous determination of medication components in pharmaceutical samples using ordinary methods have some difficulties and therefore these determinations usually were made by expensive methods and instruments. Chemometric methods are an effective way to analyze several components simultaneously.

Objective::

In this paper, a novel approach based on Bayesian regularized artificial neural network is developed for the determination of Loratadine, Naproxen, and Diclofenac in water using UV-Vis spectroscopy.

Methods:

A dataset is collected by performing several chemical experiments and recording the UV-Vis spectra and actual constituent values. The effect of a different number of neurons in the hidden layer was analyzed based on final mean square error, and the optimum number was selected. Principle Component Analysis (PCA) was also applied to the data. Other back-propagation methods, such as Levenberg-Marquardt, scaled conjugate gradient, and resilient backpropagation, were tested.

Results::

In order to see the proposed network performance, it was performed on two crossvalidation methods, namely partitioning data into train and test parts, and leave-one-out technique. Mean square errors between expected results and predicted ones implied that the proposed method has a strong ability in predicting the expected values.

Conclusion::

he results showed that the Bayesian regularization algorithm has the best performance among other methods for simultaneous determination of Loratadine, Naproxen, and Diclofenac in water samples.

Designing and understanding light-harvesting devices with machine learning

Understanding the fundamental processes of light-harvesting is crucial to the development of clean energy materials and devices. Biological organisms have evolved complex metabolic mechanisms to efficiently convert sunlight into chemical energy. Unraveling the secrets of this conversion has inspired the design of clean energy technologies, including solar cells and photocatalytic water splitting. Describing the emergence of macroscopic properties from microscopic processes poses the challenge to bridge length and time scales of several orders of magnitude. Machine learning experiences increased popularity as a tool to bridge the gap between multi-level theoretical models and Edisonian trial-and-error approaches. Machine learning offers opportunities to gain detailed scientific insights into the underlying principles governing light-harvesting phenomena and can accelerate the fabrication of light-harvesting devices.

The Sample, the Spectra and the Maths-The Critical Pillars in the Development of Robust and Sound Applications of Vibrational Spectroscopy

The last two decades have witnessed an increasing interest in the use of the so-called rapid analytical methods or high throughput techniques. Most of these applications reported the use of vibrational spectroscopy methods (near infrared (NIR), mid infrared (MIR), and Raman) in a wide range of samples (e.g., food ingredients and natural products). In these applications, the analytical method is integrated with a wide range of multivariate data analysis (MVA) techniques (e.g., pattern recognition, modelling techniques, calibration, etc.) to develop the target application. The availability of modern and inexpensive instrumentation together with the access to easy to use software is determining a steady growth in the number of uses of these technologies. This paper underlines and briefly discusses the three critical pillars-the sample (e.g., sampling, variability, etc.), the spectra and the mathematics (e.g., algorithms, pre-processing, data interpretation, etc.)-that support the development and implementation of vibrational spectroscopy applications.

Monitoring the authenticity of pu'er tea via chemometric analysis of multielements and stable isotopes

Mineral elements and stable isotopes combined with stoichiometric methods were used as a potential tool for first authenticating Chinese tea according to it's production year. A total of 86 mineral elements and stable isotope compositions were determined from the Xiangzhujing Pu'er tea in five different production years using ICP-MS and ICP-OES. On the basis of 78 statistically significant mineral elements and stable isotopes, HCA, PCA, PLS-DA, BP-ANN, and LDA were employed to build authentication models for predicting the Pu'er tea with different production years. The clustering results of the HCA and PCA were worse than that of PLS-DA, BP-ANN, and LDA. The PLS-DA model displayed a perfect model performance (R²X = 0.86, R²Y = 0.974, and Q² = 0.922). The authentication performance of LDA and BP-ANN revealed their 100% recognition sensitivity and prediction ability and was thus better than that of PLS-DA. Mn, ⁶⁸Zn, and ²⁰³Tl were the markers for enabling the successful authentication of Pu'er tea with different production years. This study contributes toward generalizing the use of mineral element and stable isotope fingerprinting combined with LDA and BP-ANN as a promising tool for authentication of tea worldwide.

Improved geographical origin discrimination for tea using ICP-MS and ICP-OES techniques in combination with chemometric approach

BACKGROUND

There is an urgent need to strengthen the testing and certification of geographically iconic foods, as well as to use discriminatory science and technology for their regulation and verification. Multi-element and stable isotope analyses were combined to provide a new chemometric approach for improving the discrimination tea samples from different geographical origins. Different stoichiometric methods [principal component analysis (PCA), hierarchical cluster analysis (HCA), partial least squares-discriminant analysis (PLS-DA), back propagation based artificial neural network (BP-ANN) and linear discriminant analysis (LDA)] were used to demonstrate this discrimination approach using Yongchuanxiuya tea samples in an experimental test.

RESULTS

Multi-element and stable isotope analyses of tea samples using inductively coupled plasma mass spectrometry and inductively coupled plasma optical emission spectrometry easily distinguished the geographical origins. However, the clustering ability of the two unsupervised learning methods (PCA and HCA) were worse compared to that of the three supervised learning methods (PLS-DA, BP-ANN and LDA). BP-ANN and LDA, with 100% recognition and prediction abilities, were found to be better than PLS-DA. ⁸⁶ Sr and ¹¹² Cd were the markers enabling the successful classification of tea samples according to their geographical origins. Under the validation by 'blind' dataset, the prediction accuracies of the BP-ANN and LDA methods were all greater than 90%. The LDA method showed the best performance, with an accuracy of 100%.

CONCLUSION

In summary, determination of mineral elements and stable isotopes using inductively coupled plasma mass spectrometry and inductively coupled plasma optical emission spectrometry techniques coupled with chemometric methods, especially the LDA method, is a good approach for improving the authentication of a diverse range of tea. The present study contributes toward generalizing the use of fingerprinting mineral elements and stable isotopes as a promising tool for testing the geographic roots of tea and food worldwide. © 2020 Society of Chemical Industry.

The Use of Artificial Neural Networks to Analyze Greenhouse Gas and Air Pollutant Emissions from the Mining and Quarrying Sector in the European Union

The European Union (EU) is considered one of the most economically developed regions worldwide. It was driven by the mining industry for several decades. Despite certain changes in this area, a number of mineral and energy resources are still being mined in the EU. Nevertheless, mining activities are accompanied by many unfavorable phenomena, especially for the environment, such as greenhouse gas and air pollutant emissions. The great diversity of the EU countries in terms of the size of the “mining and quarrying” sector means that both the volume and structure of these emissions in individual countries varies. In order to assess the current state of affairs, research was conducted to look at the structure and volume of these emissions in individual EU countries. The aim of the study was to divide these countries into homogenous groups by structure and volume of studied emissions. In order to reflect both the specificity and diversity of the EU countries, this division was based on the seven most important gases (CO2, CH4, N2O, NH3, NMVOC, CO, NOx) and two types of particulate matter (PM 2.5, PM 10) emitted into the atmosphere from the sector in question. The volume of studied emissions was also compared to the number of inhabitants of each EU country and the gross value added (GVA) by the mining and quarrying sector. This approach enabled a new and broader view on the issue of gas and air pollutant emissions associated with mining activities. The artificial Kohonen’s neural networks were used for the analysis. The developed method, the analyses and the results constitute a new approach to studying such emissions in the EU. Research that looks only at the emission of harmful substances into the environment in relation to their absolute values fail to fully reflect the complexity of this problem in individual EU countries. The presented approach and the results should broaden the knowledge in the field of harmful substance emissions from the mining and quarrying sector, which should be utilized in the process of implementing the new European climate strategy referred to as “The European Green Deal”.

Analysis on biological importance of antiseptic drug, O-Benzyl hydroxylamine, by the application of spectroscopic and theoretical tools

Biological importance of antiseptic drug; O-Benzyl hydroxylamine was explored using QSAR studies for ultimate usage for treating fungal infections. In this research work, the molecular spectroscopic tool and theoretical calculation method of analysis. The data acquired from both tools were evaluated and compared to validate structural and vibrational characteristics. Mulliken charge displacement around molecular site in order for exploring electronic properties to find out the cause of inducement of drug potential. The Lipinski rule of five was evaluated for the measurement of biological importance of the drug compound. The lipophilicity and topological surface area of the drug was monitored for determining biological process activity. The partial involvement of compositional bonds of the molecule was appraised for influential vibrational characteristics. The chemical environment for making chemical property was monitored from the uniform and asymmetrical chemical shift of core and allied carbons. The resultant oscillating potential orientation in the molecular site was identified and the residing zones were recognized to find out the origin of drug potential. The occurrence of CT complex process was studied and the CTC was found to be CC and C–N for generating drug activeness. The enhancement of hyper active polarization was measured in first and second order from which the charge level pulling on different entities were observed for ensuring the biological affinity of the compound. The enantiomer characteristics were thoroughly studied to measure the level of toxicity.

UHPLC Analysis of Saffron (Crocus sativus L.): Optimization of Separation Using Chemometrics and Detection of Minor Crocetin Esters

Ultra-high performance liquid chromatography (UHPLC) coupled with diode array detection (DAD) was applied to improve separation and detection of mono- and bis-glucosyl esters of crocetin (crocins), the main red-colored constituents of saffron (Crocus sativus L.), and other polar components. Response surface methodology (RSM) was used to optimise the chromatographic resolution on the Kinetex C18 (Phenomenex) column taking into account of the combined effect of the column temperature, the eluent flow rate and the slope of a linear eluent concentration gradient. A three-level full-factorial design of experiments was adopted to identify suitable combinations of the above factors. The influence of the separation conditions on the resolutions of 22 adjacent peaks was simultaneously modelled by a multi-layer artificial neural network (ANN) in which a bit string representation was used to identify the target analytes. The chromatogram collected under the optimal separation conditions revealed a higher number of crocetin esters than those already characterised by means of mass-spectrometry data and usually detected by HPLC. Ultra-high performance liquid chromatography analyses carried out on the novel Luna Omega Polar C18 (Phenomenex) column confirmed the large number of crocetin derivatives. Further work is in progress to acquire mass-spectrometry data and to clarify the chemical structure to the newly found saffron components.

Generating Focused Molecule Libraries for Drug Discovery with Recurrent Neural Networks

In de novo drug design, computational strategies are used to generate novel molecules with good affinity to the desired biological target. In this work, we show that recurrent neural networks can be trained as generative models for molecular structures, similar to statistical language models in natural language processing. We demonstrate that the properties of the generated molecules correlate very well with the properties of the molecules used to train the model. In order to enrich libraries with molecules active toward a given biological target, we propose to fine-tune the model with small sets of molecules, which are known to be active against that target. Against Staphylococcus aureus, the model reproduced 14% of 6051 hold-out test molecules that medicinal chemists designed, whereas against Plasmodium falciparum (Malaria), it reproduced 28% of 1240 test molecules. When coupled with a scoring function, our model can perform the complete de novo drug design cycle to generate large sets of novel molecules for drug discovery.

Automating drug discovery

Small-molecule drug discovery can be viewed as a challenging multidimensional problem in which various characteristics of compounds - including efficacy, pharmacokinetics and safety - need to be optimized in parallel to provide drug candidates. Recent advances in areas such as microfluidics-assisted chemical synthesis and biological testing, as well as artificial intelligence systems that improve a design hypothesis through feedback analysis, are now providing a basis for the introduction of greater automation into aspects of this process. This could potentially accelerate time frames for compound discovery and optimization and enable more effective searches of chemical space. However, such approaches also raise considerable conceptual, technical and organizational challenges, as well as scepticism about the current hype around them. This article aims to identify the approaches and technologies that could be implemented robustly by medicinal chemists in the near future and to critically analyse the opportunities and challenges for their more widespread application.

Chemometrics Methods for Specificity, Authenticity and Traceability Analysis of Olive Oils: Principles, Classifications and Applications

Background. Olive oils (OOs) show high chemical variability due to several factors of genetic, environmental and anthropic types. Genetic and environmental factors are responsible for natural compositions and polymorphic diversification resulting in different varietal patterns and phenotypes. Anthropic factors, however, are at the origin of different blends’ preparation leading to normative, labelled or adulterated commercial products. Control of complex OO samples requires their (i) characterization by specific markers; (ii) authentication by fingerprint patterns; and (iii) monitoring by traceability analysis. Methods. These quality control and management aims require the use of several multivariate statistical tools: specificity highlighting requires ordination methods; authentication checking calls for classification and pattern recognition methods; traceability analysis implies the use of network-based approaches able to separate or extract mixed information and memorized signals from complex matrices. Results. This chapter presents a review of different chemometrics methods applied for the control of OO variability from metabolic and physical-chemical measured characteristics. The different chemometrics methods are illustrated by different study cases on monovarietal and blended OO originated from different countries. Conclusion. Chemometrics tools offer multiple ways for quantitative evaluations and qualitative control of complex chemical variability of OO in relation to several intrinsic and extrinsic factors.

Cultivar classification of Apulian olive oils: Use of artificial neural networks for comparing NMR, NIR and merceological data

The development of an efficient and accurate method for extra-virgin olive oils cultivar and origin authentication is complicated by the broad range of variables (e.g., multiplicity of varieties, pedo-climatic aspects, production and storage conditions) influencing their properties. In this study, artificial neural networks (ANNs) were applied on several analytical datasets, namely standard merceological parameters, near-infra red data and ¹H nuclear magnetic resonance (NMR) fingerprints, obtained on mono-cultivar olive oils of four representative Apulian varieties (Coratina, Ogliarola, Cima di Mola, Peranzana). We analyzed 888 samples produced at a laboratory-scale during two crop years from 444 plants, whose variety was genetically ascertained, and on 17 industrially produced samples. ANN models based on NMR data showed the highest capability to classify cultivars (in some cases, accuracy>99%), independently on the olive oil production process and year; hence, the NMR data resulted to be the most informative variables about the cultivars.

Investigating the discrimination potential of linear and nonlinear spectral multivariate calibrations for analysis of phenolic compounds in their binary and ternary mixtures and calculation pKa values

Vanillin (VA), vanillic acid (VAI) and syringaldehyde (SIA) are important food additives as flavor enhancers. The current study for the first time is devote to the application of partial least square (PLS-1), partial robust M-regression (PRM) and feed forward neural networks (FFNNs) as linear and nonlinear chemometric methods for the simultaneous detection of binary and ternary mixtures of VA, VAI and SIA using data extracted directly from UV-spectra with overlapped peaks of individual analytes. Under the optimum experimental conditions, for each compound a linear calibration was obtained in the concentration range of 0.61-20.99 [LOD=0.12], 0.67-23.19 [LOD=0.13] and 0.73-25.12 [LOD=0.15] μgmL(-1) for VA, VAI and SIA, respectively. Four calibration sets of standard samples were designed by combination of a full and fractional factorial designs with the use of the seven and three levels for each factor for binary and ternary mixtures, respectively. The results of this study reveal that both the methods of PLS-1 and PRM are similar in terms of predict ability each binary mixtures. The resolution of ternary mixture has been accomplished by FFNNs. Multivariate curve resolution-alternating least squares (MCR-ALS) was applied for the description of spectra from the acid-base titration systems each individual compound, i.e. the resolution of the complex overlapping spectra as well as to interpret the extracted spectral and concentration profiles of any pure chemical species identified. Evolving factor analysis (EFA) and singular value decomposition (SVD) were used to distinguish the number of chemical species. Subsequently, their corresponding dissociation constants were derived. Finally, FFNNs has been used to detection active compounds in real and spiked water samples.

Multivariate statistical analysis methods in QSAR

The emphasis of this review is particularly on multivariate statistical methods currently used in quantitative structure–activity relationship (QSAR) studies.

Representing potential energy surfaces by high-dimensional neural network potentials

The development of interatomic potentials employing artificial neural networks has seen tremendous progress in recent years. While until recently the applicability of neural network potentials (NNPs) has been restricted to low-dimensional systems, this limitation has now been overcome and high-dimensional NNPs can be used in large-scale molecular dynamics simulations of thousands of atoms. NNPs are constructed by adjusting a set of parameters using data from electronic structure calculations, and in many cases energies and forces can be obtained with very high accuracy. Therefore, NNP-based simulation results are often very close to those gained by a direct application of first-principles methods. In this review, the basic methodology of high-dimensional NNPs will be presented with a special focus on the scope and the remaining limitations of this approach. The development of NNPs requires substantial computational effort as typically thousands of reference calculations are required. Still, if the problem to be studied involves very large systems or long simulation times this overhead is regained quickly. Further, the method is still limited to systems containing about three or four chemical elements due to the rapidly increasing complexity of the configuration space, although many atoms of each species can be present. Due to the ability of NNPs to describe even extremely complex atomic configurations with excellent accuracy irrespective of the nature of the atomic interactions, they represent a general and therefore widely applicable technique, e.g. for addressing problems in materials science, for investigating properties of interfaces, and for studying solvation processes.

Chemometric determination of the botanical origin for Chinese honeys on the basis of mineral elements determined by ICP-MS

In this work, the potential of mineral elements and chemometric methods as a tool to classify Chinese honeys according to their botanical origin was examined. Twelve mineral elements (Na(23), Mg(24), P(31), K(39), Ca(43), Mn(55), Fe(56), Cu(63), Zn(66), Rb(85), Sr(88), and Ba(137)) of 163 Chinese honey samples, including linden, vitex, rape, and acacia, collected from Heilongjiang, Beijing, Hebei, and Shaanxi, China, in 2013 were determined by the ICP-MS method. Principal component analysis (PCA) reduced 10 variables to four principal components and could explain 93.06% of the total variance. Partial least-squares discriminant analysis (PLS-DA) and back-propagation artificial neural network (BP-ANN) were explored to construct a classification model. By PLS-DA, the total correct classification rates for model training and cross-validation were 90.9 and 88.4%, respectively. By BP-ANN, the total correct classification rates for model training and cross-validation were 100 and 92.6%, respectively. The performance of BP-ANN was better than that of PLS-DA. The validation of the developed BP-ANN model was tested by the independent test set of 42 honey samples. Linden, vitex, and rape honey samples were predicted with an accuracy of 100%, whereas one acacia honey was predicted as rape honey with an accuracy of 92.3%. It is concluded that the profiles of mineral elements by ICP-MS with chemometric methods could be a potential and powerful tool for the classification of Chinese honey samples from different botanical origins.

PERCEPTRON ARTIFICIAL NEURAL NETWORK AND PREDICTION OF BUBBLE POINTS OF TERNARY MIXTURES CONTAINING IONIC LIQUIDS

The feasibility of using cascade neural networks to correlate the bubble points of ternary mixtures containing ionic liquids (ILs) and classical solvents was investigated. The systems investigated consisted of the ILs from the alkylimidazolium family with either the chloride, tetrafluoroborate, methylsulfate, or ethyl sulfate anions. The classical solvents investigated included 1-propanol, 2-propanol, ethanol, ethyl ethanoate, and water. A total of 272 bubble points were used in the training (205 data points) and testing (67 data points) stages. The optimized network comprised of nine neurons in the hidden layer and used the tangent-sigmoid and purelin functions as the activation functions in the hidden and output layers, respectively. This proposed network was able to correlate the ternary bubble points of both the training (AARD % = 0.13% and R2 = 0.9921) and the testing (AARD % = 0.15% and R2 = 0.9873) subsets with good accuracy, revealing the good correlative capability of the network. The statistical error analysis of all the data indicated an overall absolute average relative deviation (AARD %), mean square error (MSE) and correlation coefficient (R2) of 0.13%, 0.44% and 0.9909%, respectively.

Investigating the role of weight update functions in developing artificial neural network modeling of retention times of furan and phenol derivatives

A quantitative structure−retention relationship study has been carried out on the retention times of 63 furan and phenol derivatives using artificial neural networks (ANNs). First, a large number of descriptors were calculated using HyperChem, Mopac, and Dragon softwares. Then, a suitable number of these descriptors were selected using a multiple linear regression technique. This paper focuses on investigating the role of weight update functions in developing ANNs. Therefore, selected descriptors were used as inputs for ANNs with six different weight update functions including the Levenberg−Marquardt back-propagation network, scaled conjugate gradient back-propagation network, conjugate gradient back-propagation with Powell−Beale restarts network, one-step secant back-propagation network, resilient back-propagation network, and gradient descent with momentum back-propagation network. Comparison of the results indicates that the Levenberg−Marquardt back-propagation network has better predictive power than the other methods.

Artificial neural network processing of stripping analysis responses for identifying and quantifying heavy metals in the presence of intermetallic compound formation

Feed-forward neural networks have been trained to identify and quantify heavy metals in mixtures under conditions where there were significant complications due to intermetallic compound formation. The networks were shown to be capable of (i) correlating voltammetric responses with individual heavy metals in complex mixtures, (ii) determining the relationship between responses and concentrations (including nonlinear relationships due to overlapping peaks and intermetallic compound formation), and (iii) rapidly determining concentrations of individual components from mixtures once trained. Using simulated data, modeled after complex interactions experimentally observed in samples containing Cu and Zn, it has been demonstrated that networks containing two layers of neurons (a nonlinear hidden layer and a linear output layer) can be trained to calculate concentrations under a variety of complicated situations. These include, but are not limited to, cases where the response of the intermetallic compound formed is observed as a shoulder of one of the pure metals and cases where the response of the intermetallic compound formed is not observed in the potential window. In addition, the network described above was trained to simultaneously determine concentrations of four metals (Cu, Pb, Cd, and Zn) in a concentration range where all responses were complicated by intermetallic compound formation (1-500 ppb).