Application of neural networks in analysis of thermal decomposition of CoSO4·7H2O

Thermal decomposition of CoSO₄·7H₂O was investigated by simultaneous DTA-TG techniques and XRD method. Neural networks were used for DTA-TG curves analysis. Additionally, the network architecture (GRNN - Generalized Regression Neural Networks) and its statistical parameters were calculated. This method permits to generate DTA-TG curves without using kinetic models. This revised version was published online in July 2006 with corrections to the Cover Date.     

Comparison of supervised pattern recognition methods with McNemar’s statistical test: Application to qualitative analysis of sugar beet by near-infrared spectroscopy

The application of supervised pattern recognition methodology is becoming important within chemistry. The aim of the study is to compare classification method accuracies by the use of a McNemar’s statistical test. Three qualitative parameters of sugar beet are studied: disease resistance (DR), geographical origins and crop periods. Samples are analyzed by near-infrared spectroscopy (NIRS) and by wet chemical analysis (WCA). Firstly, the performances of eight well-known classification methods on NIRS data are compared: Linear Discriminant Analysis (LDA), K-Nearest Neighbors (KNN) method, Soft Independent Modeling of Class Analogy (SIMCA), Discriminant Partial Least Squares (DPLS), Procrustes Discriminant Analysis (PDA), Classification And Regression Tree (CART), Probabilistic Neural Network (PNN) and Learning Vector Quantization (LVQ) neural network are computed. Among the three data sets, SIMCA, DPLS and PDA have the highest classification accuracies. LDA and KNN are not significantly different. The non-linear neural methods give the less accurate results. The three most accurate methods are linear, non-parametric and based on modeling methods. Secondly, we want to emphasize the power of near-infrared reflectance data for sample discrimination. McNemar’s tests compare classification developed with WCA or with NIRS data. For two of the three data sets, the classification results are significantly improved by the use of NIRS data.     

Boron Nitride Nanomaterials for Thermal Management Applications

Hexagonal boron nitride nanosheets (BNNs) are analogous to their two‐dimensional carbon counterparts in many materials properties, in particular, ultrahigh thermal conductivity, but also offer some unique attributes, including being electrically insulating, high thermal stability, chemical and oxidation resistance, low color, and high mechanical strength. Significant recent advances in the production of BNNs, understanding of their properties, and the development of polymeric nanocomposites with BNNs for thermally conductive yet electrically insulating materials and systems are highlighted herein. Major opportunities and challenges for further studies in this rapidly advancing field are also discussed.     

Recursive neural networks prediction of glass transition temperature from monomer structure: an application to acrylic and methacrylic polymers

We propose a new method based on a Recursive Neural Network (RecNN) for predicting polymer properties from their structured molecular representations. RecNN allows for a completely novel approach to QSPR analysis by direct adaptive processing of molecular graphs. This model joins the representational power of structured domains with Neural Network ability to capture underlying complex relationships in the data by a process of training from examples. To this aim, a structured representation was designed for the modelling of polymer structures. The adopted representation can account also for average macromolecule characteristics, such as degree of polymerization, stereoregularity, comonomer distribution. To begin with, this model was applied to the prediction of the glass transition temperature of (meth)acrylic polymers with different degree of main chain tacticity. The results so far obtained indicate that the proposed representation of polymer structure can convey information on both the repeating unit structure and average polymer features. The ability of the proposed RecNN method of treating this structured representation makes this method more general and flexible with respect to standard literature methods. Moreover, the same model can handle at the same time the Tg of polymer samples present in only one tacticity form together with that of polymer with different stereoregularity.     

机器学习方法用于二氢叶酸还原酶抑制剂的活性预测

分别采用支持向量学习机、人工神经网络、调节性逻辑回归和K-最临近等机器学习方法对761个二氢叶酸还原酶抑制剂建立了其活性分类预测模型. 采用组成描述符和拓扑描述符表征抑制剂的分子结构及物理化学性质, 使用Kennard-Stone方法进行训练集的设计, 并用Metropolis Monte Carlo模拟退火方法作变量选择. 结果表明, 支持向量学习机优于其它机器学习方法, 所得到的最优模型具有较好的预测结果, 其预测正确率为91.62%. 说明通过合适的训练集设计及变量选择, 支持向量学习机方法可以很好地用于二氢叶酸还原酶抑制剂的活性分类预测.     

Temperature dependence of the crystal lattice organization of coordination compounds involving nitronyl nitroxide radicals: a magnetic and structural investigation

Four new mononuclear complexes of formula Cd(PN)(4)(NCS)(2) (A), Cd(PNN)(4)(N(3))(2) (B), Zn(PNN)(4)(N(3))(2) (C), and Zn(PNN)(2)(NCS)(2) (D), where PNN stands for 2-(4-pyridyl)-4,4,5,5-tetramethylimidazoline-1-oxyl-3-oxide and PN for 2-(4-pyridyl)-4,4,5,5-tetramethylimidazoline-1-oxyl, were synthesized and structurally and magnetically characterized. The X-ray structures of compounds B and C were also determined at 90 K. Compounds A[bond]C crystallize in the triclinic space group P 1 macro (No. 2), and D crystallizes in the monoclinic space group P2(1)/m (No. 11). A[bond]C adopt a centrosymmetric distorted octahedral geometry in which the metal ions are bonded to four radical ligands through the nitrogen atom of the pyridyl rings and the azido or thiocyanato ligands occupy the apical positions. Compound D adopts a distorted tetrahedral geometry in which the zinc ion is bonded to two radicals and two thiocyanato ligands. As suggested by their magnetic behavior, the low-temperature X-ray structures of B and C show that these compounds undergo a clear structural change with respect to the room-temperature structures. The experimental magnetic behaviors were perfectly reproduced by a dimer model for A[bond]C and an alternating chain model for D while the sudden breaks observed in the chi(M)T versus T curves for B and C were well accounted for by the high- and low-temperature X-ray structures. For all these complexes the crystal structures favor significant overlap between molecular magnetic orbitals leading to rather strong intermolecular antiferromagnetic interactions.     

人工神经网络用于连钱草中总黄酮和三萜酸类成分的含量预测

以连钱草的毛细管电泳指纹图谱为输入数据,以总黄酮和三萜酸类成分含量为输出数据,构建了反向传播网络、径向基函数网络和广义回归网络三种人工神经网络模型.采用三种网络模型和两种预测方法对未知样本的总黄酮和三萜酸类成分含量进行了预测,并分别比较了三种网络和两种预测方法的预测结果.另外,结合聚类分析结果和输入数据的相似度,分析了预测误差的来源.结果表明:三种网络对大部分样本的预测值与实际值都比较接近,而广义回归网络的预测效果最优;扣除奇异值后,广义回归网络的两种预测方法对未知样本的总黄酮和三萜酸类成分含量的平均预测误差分别为10.9%和0.00073%.     

基于布谷鸟算法与BP神经网络的煤灰变形温度预测

以120种煤样为数据基础,采用布谷鸟算法(CS)优化BP(Back Propagation)神经网络,建立了CSBP模型对单煤、煤掺添加剂和配煤等3类样本的煤灰变形温度(DT)样本进行预测。模型以煤灰化学成分及其组合参数等13个变量作为输入量,以变形温度(DT)作为输出量。CSBP模型预测结果与BP神经网络模型预测结果进行对比发现,无论是单煤、煤掺添加剂还是配煤,CSBP模型较BP模型对煤灰变形温度(DT)的预测都更加精准,平均相对误差分别达到了3.11%、4.08%和4.22%。另外,对比3类样本预测结果发现,无论是CSBP模型还是BP模型,相比单煤预测而言,煤掺添加剂及配煤的预测误差都有明显的增加。     

Network visualization system for computational chemistry

Network Visualization System for Computational Chemistry (NVSCC) is a molecular graphics program designed for the visualization of molecular assemblies. NVSCC accepts the output files from the most popular ab initio quantum chemical programs, GAUSSIAN and GAMESS, and provides visualization of molecular structures based on atomic coordinates. The main differences between NVSCC and other programs are: Network support due to built-in FTP and telnet clients, which allows for the processing of output from and the sending of input to different computer systems and operating systems. The possibility of working with output files in real time mode. The possibility of animation from an output file during all steps of optimization. The quick processing of huge volumes of data. The development of custom interfaces.     

Solubility enhancement of decitabine as anticancer drug via green chemistry solvent: Novel computational prediction and optimization

Nowadays, supercritical fluid technology (SFT) has been an interesting scientific subject in disparate industrial-based activities such as drug delivery, chromatography, and purification. In this technology, solubility plays an incontrovertible role. Therefore, achieving more knowledge about the development of promising numerical/computational methods of solubility prediction to validate the experimental data may be advantageous for increasing the quality of research and therefore, the efficacy of novel drugs. Decitabine with the chemical formula C₈H₁₂N₄O₄ is a chemotherapeutic agent applied for the treatment of disparate bone-marrow-related malignancies such as acute myeloid leukemia (AML) by preventing DNA methyltransferase and activation of silent genes. This study aims to predict the optimum value of decitabine solubility in CO₂SCF by employing different machine learning-based mathematical models. In this investigation, we used AdaBoost (Adaptive Boosting) to boost three base models such as Linear Regression (LR), Decision Tree (DT), and GRNN. We used a dataset that has 32 sample points to make solubility models. One of the two input features is P (bar) and the other is T (k). ADA-DT (Adaboost Algorithm-Decision Tree), ADA-LR (Adaboost Algorithm-Linear Regresion), and ADA-GRNN (Generative Regression Neural Network) models showed MAE of 6.54 × 10^?5, 4.66 × 10^?5, and 8.35 × 10^?5, respectively. Also, in terms of R-squared score, these models have 0.986, 0.983, and 0.911 scores, respectively. ADA-LR was selected as the primary model according to numerical and visual analysis. Finally, the optimal values are (P = 400 bar, T = 3.38 K × 102, Y = 1.064 × 10^?3 mol fraction) using this model.     

Optimization of metformin HCl 500 mg sustained release matrix tablets using Artificial Neural Network (ANN) based on Multilayer Perceptrons (MLP) model

The aim of the present study was to apply the simultaneous optimization method incorporating Artificial Neural Network (ANN) using Multi-layer Perceptron (MLP) model to the development of a metformin HCl 500 mg sustained release matrix tablets with an optimized in vitro release profile. The amounts of HPMC K15M and PVP K30 at three levels (-1, 0, +1) for each were selected as casual factors. In vitro dissolution time profiles at four different sampling times (1 h, 2 h, 4 h and 8 h) were chosen as output variables. 13 kinds of metformin matrix tablets were prepared according to a 2(3) factorial design (central composite) with five extra center points, and their dissolution tests were performed. Commercially available STATISTICA Neural Network software (Stat Soft, Inc., Tulsa, OK, U.S.A.) was used throughout the study. The training process of MLP was completed until a satisfactory value of root square mean (RSM) for the test data was obtained using feed forward back propagation method. The root mean square value for the trained network was 0.000097, which indicated that the optimal MLP model was reached. The optimal tablet formulation based on some predetermined release criteria predicted by MLP was 336 mg of HPMC K15M and 130 mg of PVP K30. Calculated difference (f(1) 2.19) and similarity (f(2) 89.79) factors indicated that there was no difference between predicted and experimentally observed drug release profiles for the optimal formulation. This work illustrates the potential for an artificial neural network with MLP, to assist in development of sustained release dosage forms.     

Nickel complexes incorporating an amido phosphine chelate with a pendant amine arm: synthesis, structure, and catalytic Kumada coupling

A series of organonickel(II) complexes incorporating an amido phosphine ligand tethered with an amino pendant have been prepared and characterized. Deprotonation of N-(dimethylaminoethyl)-2-diphenylphosphinoaniline (H[PNN]) with one equivalent of n-BuLi in ethereal or hydrocarbon solutions at -35 °C generates cleanly dimeric {Li[PNN]}(2) as yellow crystals. The reaction of NiCl(2)(DME) with {Li[PNN]}(2) in THF at -35 °C affords green crystalline [PNN]NiCl. Treating [PNN]NiCl with NaX in acetone solutions gives [PNN]NiX (X = Br, I). Alkylation or arylation of [PNN]NiCl with appropriate Grignard reagents in THF at -35 °C produces red crystalline [PNN]NiR (R = Me, Et, i-Bu, n-hexyl, CH(2)Ph, Ph). The chloride complex [PNN]NiCl was found to be an active catalyst precursor for Kumada coupling reactions of PhX (X = I, Br, Cl) with aryl or alkyl Grignard reagents, including those containing β-hydrogen atoms. The X-ray structures of {Li[PNN]}(2) and [PNN]NiX (X = Cl, Br, Me, Et, n-hexyl) are reported.