基于GA_XGBoost模型的大学生科研能力预测问题研究

科学评价大学生科研创新能力对我国科研水平的提高具有重要意义.采用机器学习模型来预测大学生科研能力可以起到良好的效果,提出一种GAXGBoost模型来实现对大学生的科研能力预测.此模型是以Xgboost算法为基础,然后充分利用遗传算法的全局搜索能力自动搜索Xgboost最优超参数,避免了人为经验调参不准确的缺陷,最后采用精英选择策略以此确保每一轮都是最佳的进化结果.通过分析表明,所采用的GAXGBoost模型在大学生科研能力预测的结果中具有很高的精度,将此模型与Logistic Regression、Random Forest、SVM等模型进行对比,GAXGBoost模型的预测精度最高.     

游客情感计算的文本大数据挖掘方法比较研究

旅游文本大数据以其方便、快捷和低门槛的特点为游客情感计算提供了极大便利，已经成为旅游大数据的主要来源之一。基于大数据理论和情感理论，以文本大数据为数据源，在全面梳理国内外情感计算相关成果的基础上，利用人工智能中的逻辑/算法编程方法、机器学习方法、深度学习方法对旅游文本大数据进行挖掘，探索最佳的基于文本大数据的游客情感计算方法。研究发现：（1）基于情感词典的游客情感计算模型，其核心是构建情感词典和设计情感计算规则，方法简单，容易实现，适用语料范围广。（2）机器学习，用统计学方法抽取文本中的特征项，具有非线性特征，可靠性较线性特征的情感词典方法高。（3）基于深度学习技术的游客情感计算，效果良好，准确率在85%以上。训练多领域的文本语料易于移植，实用性强，且泛化能力好，较适合大数据时代游客情感计算研究。     

桉树杂交种与其亲本的近红外光谱判别

研究桉树控制授粉后目标性状的基因作用方式是探索其基因重组规律的重要内容。常规的数量统计分析精度往往不高,而DNA分析的专业要求高,且费时费力。该研究利用近红外光谱(NIRs)研究不同基因型桉树杂交种、亲本及杂交种与亲本间近红外光谱信息的关系,探索NIRs用于桉树杂交种与其亲本判别的可行性和准确性。以控制授粉的桉树亲本及其杂交F1代材料为对象,每种基因型从各自田间试验分别选取10个单株,采集树冠中上部新鲜健康叶片。用手持式近红外仪Phazir Rx(1624)采集桉树杂交种与其亲本叶片的NIRs信息。每单株选10片完全生理成熟的健康叶片,避开叶脉扫描其正面光谱5次,以50条NIRs信息的均值代表单个叶片的NIRs信息,最终每个基因型获得10条NIRs信息。对原始NIRs采用二阶多项式S.G一阶导数预处理。预处理后的NIRs用于多元统计分析,首先对桉树杂交亲本和子代样本进行主成分分析(PCA),直观展示不同基因型的分类情况。然后运用簇类独立软模式(SIMCA)和偏最小二乘判别分析(PLS-DA)两种有监督的判别模式验证NIRs用于桉树杂交种与其亲本树种的分类判别效果。PCA结果显示,不同的亲本间、杂交种间及杂交种与亲本间样本的主因子得分可以清晰地将各基因型分开。SIMCA模式判别分析中,桉树杂交种样本到亲本PCA模型的样本距离显示,待判别样本能够形成单独的聚类,且能直观反映两者的遗传相似。PLS-DA判别结果显示,桉树杂交亲本的PLS模型能通过预测其杂交子代的响应变量将其与亲本准确分开。结果表明,桉树叶片的NIRs信息可以准确地反映桉树杂交子代遗传信息的传递规律,NIRs判别模型可以准确地将各种基因型予以区分。因此,NIRs信息不仅可用于桉树杂交种和纯种的定性判别,还可以分析桉树基因重组过程中加性遗传效应的大小,从而为桉树遗传基础分析及其育种改良研究提供理论支撑。     

紫色球杆菌视紫红质光谱特性的机器学习研究

近年来，机器学习等人工智能技术被应用于蛋白质工程，其在蛋白质结构、功能预测、催化活性等研究中具有独特优势。在未知蛋白质结构的情况下，将蛋白质序列和功能特性与机器学习相结合，基于序列-活性关系（innovative sequence-activity relationship，ISAR）算法，将蛋白质氨基酸序列数字化，用快速傅里叶变换（fast four transform，FFT）进行预处理，再进行偏最小二乘回归建模，可在数据集较少情况下拟合得到最佳模型。通过机器学习对紫色球杆菌视紫红质（gloeobacter violaceus rhodopsin，GR）的突变体蛋白质氨基酸序列与光谱最大吸收波长进行建模，获得了最佳模型。用最佳索引LEVM760106建模得到的确定系数R² 为0.944，均方误差E为11.64。用小波变换进行的预处理，其R² 虽也约为0.944，但E大于11.64，不及FFT进行的预处理。方法较好地解决了蛋白质序列与功能特性之间的数学建模问题，在蛋白质工程中可为预测更优的突变体提供支持。     

What is the form of muscimol from fly agaric mushroom (Amanita muscaria) in water? An insight from NMR experiment supported by molecular modeling

The biologically active alkaloid muscimol is present in fly agaric mushroom (Amanita muscaria), and its structure and action is related to human neurotransmitter _γ-aminobutyric acid (GABA). The current study reports on determination of muscimol form present in water solution using multinuclear ¹H and ¹³C nuclear magnetic resonance (NMR) experiments supported by density functional theory molecular modeling. The structures of three forms of free muscimol molecule both in the gas phase and in the presence of water solvent, modeled by polarized continuous model, and nuclear magnetic isotropic shieldings, the corresponding chemical shifts, and indirect spin–spin coupling constants were calculated. Several J-couplings observed in proton and carbon NMR spectra, not available before, are reported. The obtained experimental spectra, supported by theoretical calculations, favor the zwitterion form of muscimol in water. This structure differs from NH isomer, previously determined in dimethyl sulfoxide (DMSO) solution. In addition, positions of signals C3 and C5 are reversed in both solvents.     

Study of Endogen Substrates,Drug Substrates and Inhibitors Binding Conformations on MRP4 and Its Variants by Molecular Docking and Molecular Dynamics

Multidrug resistance protein-4 (MRP4) belongs to the ABC transporter superfamily and promotes the transport of xenobiotics including drugs. A non-synonymous single nucleotide polymorphisms (nsSNPs) in the ABCC4 gene can promote changes in the structure and function of MRP4. In this work, the interaction of certain endogen substrates, drug substrates, and inhibitors with wild type-MRP4 (WT-MRP4) and its variants G187W and Y556C were studied to determine differences in the intermolecular interactions and affinity related to SNPs using protein threading modeling, molecular docking, all-atom, coarse grained, and umbrella sampling molecular dynamics simulations (AA-MDS and CG-MDS, respectively). The results showed that the three MRP4 structures had significantly different conformations at given sites, leading to differences in the docking scores (DS) and binding sites of three different groups of molecules. Folic acid (FA) had the highest variation in DS on G187W concerning WT-MRP4. WT-MRP4, G187W, Y556C, and FA had different conformations through 25 ns AA-MD. Umbrella sampling simulations indicated that the Y556C-FA complex was the most stable one with or without ATP. In Y556C, the cyclic adenosine monophosphate (cAMP) and ceefourin-1 binding sites are located out of the entrance of the inner cavity, which suggests that both cAMP and ceefourin-1 may not be transported. The binding site for cAMP and ceefourin-1 is quite similar and the affinity (binding energy) of ceefourin-1 to WT-MRP4, G187W, and Y556C is greater than the affinity of cAMP, which may suggest that ceefourin-1 works as a competitive inhibitor. In conclusion, the nsSNPs G187W and Y556C lead to changes in protein conformation, which modifies the ligand binding site, DS, and binding energy.     

Characterization of new Pt(IV)–thiazole complexes: Analytical,spectral, molecular modeling and molecular docking studies and applications in two opposing pathways

New thiazole derivatives were synthesized and fully characterized, then coordinated with PtCl₄ salt. Also, the newly synthesized Pt(IV) complexes were investigated analytically (elemental and thermogravimetric analyses), spectrally (infrared, UV–visible, mass, ¹H NMR, ¹³C NMR, X‐ray diffraction) as well as theoretically (kinetics, modeling and docking). The data extracted led to the establishment of the best chemical and structural forms. Octahedral geometry was the only formula proposed for all complexes, which is favorable for d⁶ systems. The molecular ion peaks from mass spectral analysis coincide with all analytical data, confirming the molecular formula proposed. X‐ray diffraction (XRD) and scanning electron microscopy (SEM) allowed discrimination of features between crystalline particles and other amorphous morphology. By applying Gaussian09 as well as HyperChem 8.2 programs, the best structural forms were obtained, as well as computed significant parameters. Computed parameters such as softness, hardness, surface area and reactivity led us towards application in two opposing pathways: tumor inhibition and oxidation activation. The catalytic oxidation for CO was conducted over PtO₂, which was yielded from calcination of the most reactive complex. The success of catalytic role for synthesized PtO₂ was due to its particulate size and surface morphology, which were estimated from XRD patterns and SEM images, respectively. The antitumor activity was tested versus HCT‐116 and HepG‐2 cell lines. Mild toxicity was recorded for two of the derivatives and their corresponding complexes. This degree of toxicity is more favorable in most cases, due to exclusion of serious side effects, which is coherently attached with known antitumor drugs.     

Incremental dynamic mode decomposition: A reduced-model learner operating at the low-data limit

The present work aims at proposing a new methodology for learning reduced models from a small amount of data. It is based on the fact that discrete models, or their transfer function counterparts, have a low rank and then they can be expressed very efficiently using few terms of a tensor decomposition. An efficient procedure is proposed as well as a way for extending it to nonlinear settings while keeping limited the impact of data noise. The proposed methodology is then validated by considering a nonlinear elastic problem and constructing the model relating tractions and displacements at the observation points.     

Exploiting the diversity of time scales in the immune system: A B-cell antibody model

Using the continuous shape space formalism, we develop an immune system model involving both B lymphocytes and antibody molecules. The binding and cross-linking of receptors on B cells stimulates the cells to divide and, with a lag, to secrete antibody. Using the method of multiple scales, we show how to correctly formulate long-time-scale equations for the population dynamics of B cells, the total antibody concentration, and rate of antibody secretion. We compare our model with previous phenomenological formulations.     

机载稳瞄控制系统模型及仿真分析

瞄准线高精度稳定是机载光电稳瞄系统的主要指标和关键技术。根据四框架稳瞄系统的工作原理，以手动跟踪模式为主要研究对象，建立了稳瞄伺服控制系统模型。考虑到直升机扰动特点，对线扰动、角速率扰动、摩擦力矩、弹性力矩等各种扰动因素也建立相应的数学模型，同时在各种扰动因素作用下利用Matlab对机载高精度稳瞄系统的手动跟踪控制模式进行了仿真设计分析和理论研究，设计出适合的控制器。此模型在实际系统中获得验证，对稳瞄伺服控制系统的设计具有参考意义。