基于双向循环神经网络的汉语语音识别*

当前基于深度神经网络模型中,虽然其隐含层可设置多层,对复杂问题适应能力强,但每层之间的节点连接是相互独立的,这种结构特性导致了在语音序列中无法利用上下文相关信息来提高识别效果,而传统的循环神经网络虽然做出了改进,但是只能对上文信息进行利用。针对以上问题,该文采用可以同时利用语音序列中上下文相关信息的双向循环神经网络模型与深度神经网络模型相结合,并应用于语音识别。构建具有5层隐含层的模型,其中第3层为双向循环神经网络结构,其他层采用深度神经网络结构。实验结果表明:加入了双向循环神经网络结构的模型与其他模型相比,较好地提高了识别正确率;噪声对双向循环神经网络汉语识别有重要影响,尤其是训练集和测试集附加噪声类型不同时,单一的含噪声语音的训练模型无法适应不同噪声类型的语音识别;调整神经网络模型中隐含层神经元数量后,识别正确率并不是一直随着隐含层中神经元数量的增加而增加,神经元数量数目增加到一定程度后正确率出现了降低的趋势。     

A solvent-governed surface state strategy for rational synthesis of N and S co-doped carbon dots with multicolour fluorescence

ABSTRACT

Multicolour emissive carbon dots (CDs) are widely investigated by virtue of their merits on fluorescent properties. Method on heteroatom doping assisted with various solvents has been proved efficient in achieving multiple-colour-emissive CDs, especially long-wavelength emission. Herein, a synthesis of multicolour-emissive CDs by controlled surface function is reported. By tuning the thermal-pyrolysis temperature and molar ratio of reactants, optimal emission of the resulted CDs gradually shifts from blue to yellow light with the assistance of different solvents. According to the emissive relationship dependent on excitation, fluorescence lifetimes, and FT-IR of these CDs, the different surface states participated with S and N elements on the surface of carbogenic core govern fluorescent colours of the CDs. In terms of the applications, blue CDs (B-CDs) exhibits high sensitivity for ion detections of Ag⁺ and Fe³⁺, which is further illustrated to have different quenching mechanisms each other because that these ions have the affinity interaction with different surface groups of the CDs. Moreover, blue and yellow CDs solutions are mixed with PVP water solution to fabricate white-light CDs/PVP film, which exhibits stable fluorescence with a CIE coordinate of (0.32, 0.33) and endows these CDs as potentially fluorescent nanomaterial in the solid state lighting field.     

新型非对称二苯并[b,f][1,5]二氮杂环辛四烯的合成

以两种不同取代的2-氨基二苯甲酮为原料,氯苯为溶剂,BF3-Et2O为脱水剂,通过分子间脱水一步环化缩合制备非对称二苯并[b,f][1,5]二氮杂环辛四烯衍生物。运用HPLC监控反应过程,优化合成工艺,得到最佳反应条件为:等物质的量的两种不同取代2-氨基二苯甲酮和BF3-Et2O,在氯苯中回流反应12 h。化合物4a^4c为新化合物,其结构经1H NMR,13C NMR和MS(ESI)表征。     

A Comparative Analysis of Machine Learning Techniques for Muon Count in UHECR Extensive Air-Showers

The main goal of this work is to adapt a Physics problem to the Machine Learning (ML) domain and to compare several techniques to solve it. The problem consists of how to perform muon count from the signal registered by particle detectors which record a mix of electromagnetic and muonic signals. Finding a good solution could be a building block on future experiments. After proposing an approach to solve the problem, the experiments show a performance comparison of some popular ML models using two different hadronic models for the test data. The results show that the problem is suitable to be solved using ML as well as how critical the feature selection stage is regarding precision and model complexity.     

Dynamical Analysis of a New Chaotic Fractional Discrete-Time System and Its Control

This article proposes a new fractional-order discrete-time chaotic system, without equilibria, included two quadratic nonlinearities terms. The dynamics of this system were experimentally investigated via bifurcation diagrams and largest Lyapunov exponent. Besides, some chaotic tests such as the 0–1 test and approximate entropy (ApEn) were included to detect the performance of our numerical results. Furthermore, a valid control method of stabilization is introduced to regulate the proposed system in such a way as to force all its states to adaptively tend toward the equilibrium point at zero. All theoretical findings in this work have been verified numerically using MATLAB software package.     

Tandem Use of Optical Sensing and Machine Learning for the Determination of Absolute Configuration,Enantiomeric and Diastereomeric Ratios,and Concentration of Chiral Samples

We have developed an optical method for accurate concentration, er, and dr analysis of amino alcohols based on a simple mix‐and‐measure workflow that is fully adaptable to multiwell plate technology and microscale analysis. The conversion of the four aminoindanol stereoisomers with salicylaldehyde to the corresponding Schiff base allows analysis of the dr based on a change in the UV maximum at 420 nm that is very different for the homo‐ and heterochiral diastereomers and of the concentration of the sample using a hypsochromic shift of another absorption band around 340 nm that is independent of the analyte stereochemistry. Subsequent in situ formation of Cu^II assemblies in the absence and presence of base enables quantification of the er values for each diastereomeric pair by CD analysis. Applying a linear programming method and a parameter sweep algorithm, we determined the concentration and relative amounts of each of the four stereoisomers in 20 samples of vastly different stereoisomeric compositions with an averaged absolute percent error of 1.7 %.     

Multifunctional Sulfur-Hyperdoped Silicon Nanoparticles with Engineered Mid-Infrared Sulfur-Impurity and Free-Carrier Absorption

Si nanoparticles (NPs), which are innovative promising light-harvesting components of thin-film solar cells and key-enabling biocompatible theranostic elements of infrared-laser and radiofrequency hyperthermia-based therapies of cancer cells in tumors and metastases, are significantly advanced in their near/mid-infrared band-to-band and free-carrier absorption via donor sulfur-hyperdoping during high-throughput facile femtosecond-laser ablative production in liquid carbon disulfide. High-resolution transmission electron microscopy and Raman microscopy reveal their mixed nanocrystalline/amorphous structure, enabling the extraordinary sulfur content of a few atomic percents and very minor surface oxidation/carbonization characterized by energy-dispersive X-ray spectroscopy and X-ray photoelectron spectroscopy. A 200-nm thick layer of the nanoparticles exhibits near−mid-infrared absorbance, comparable to that of the initial 380-micron thick n-doped Si wafer (phosphor-dopant concentration ≈10¹⁵ cm⁻³), with the corresponding extinction coefficient for the hyperdoped NPs being 4–7 orders higher over the broadband spectral range of 1–25 micrometers. Such ultimate, but potentially tunable mid-IR structured, multi-band absorption of various sulfur-impurity clusters and smooth free-carrier absorption are break through advances in mid-infrared (mid-IR) laser and radiofrequency (RF) hyperthermia-based therapies, as envisioned in the RF-heating tests, and in fabrication of higher-efficiency thin-film and bulk photovoltaic devices with ultra-broad (UV−mid-IR) spectral response.     

Dinuclear gold (I)-di-N-heterocyclic carbene complexes: syntheses,structure, density functional theory calculation and photoluminescence study

The synthesis and characterizations for a series of dinuclear gold (I)-di-NHC complexes, 1–8 through the trans-metalation method of their respective silver (I)-di-NHC complexes, i–viii are reported (where NHC = N-heterocyclic carbene). The successful complexation of a series of unusual non-symmetrical and symmetrical di-NHC ligands, 3,3'-(ethane-1,2-diyl)-1-alkylbenzimidazolium-1'-butylbenzimidazolium (with alkyl = methyl, ethyl, propyl, butyl, pentyl, hexyl, heptyl, benzyl) with the gold (I) ions are suggested by elemental analysis, Fourier transform-infrared, ¹H- and ¹³C-NMR data. The ¹³C-NMR spectra of 1–8 show a singlet sharp peak in the range of 190.00–192.00 ppm, indicating the presence of a carbene carbon that bonded to the gold (I) ion. From single crystal X-ray diffraction data, the structure of complex 6 with the formula of [di-NHC-Au (I)]₂·2PF₆ is obtained [where NHC = 3,3'-(ethane-1,2-diyl)-1-hexylbenzimidazolium-1'-butylbenzimidazolium]. The photophysical study in solid state of 6 displays an intense photoluminescence with a strong emission maxima, λ_em = 480 nm, upon excitation at 340 nm at room temperature. Interestingly, the emission maximum at 77 K shows a structural character with a strong peak at 410 nm, a medium at 433 nm and a weak at 387 nm, accompanied by a tail band to about 500 nm.     

Controlled synthesis of ZnO nanoparticles from a Zn(II) coordination polymer: Structural characterization,optical properties and photocatalytic activity

A zinc coordination polymer derived from pyridine-2,6-dicarboxylate (PDC), {[Zn₂(PDC)₂]}_n, was successfully prepared via conventional, sonication and microwave-irradiation methods. The composition and characteristics of the obtained coordination polymers (CPs) were investigated by elemental analysis, TGA/DTA, X-ray diffraction and spectroscopic techniques. The so obtained CPs were heat-treated in the air at 600 °C for 2 h to produce ZnO of nanosized particles (NPs). It is of interest to note that the synthesis approach of the precursor greatly affects both the nanoparticle size and the structure of the resulting ZnO NPs. Moreover, the smallest particle size was associated with the sample derived from the ultrasonically prepared precursor. TEM analysis revealed that all samples have sphere-like morphologies. Structural analysis of the prepared ZnO samples was conducted and compared using Rietveld analysis of their PXRD patterns. Optical band gap calculations based on analysis of the UV–vis spectra of ZnO samples using Tauc's power law were achieved. The highest band gap of 3.63 eV was observed for ZnO sample obtained from the ultrasonically prepared precursor. Furthermore, the photocatalytic activity of ZnO NPs for the removal of Eosin Y color was monitored. The highest removal efficiency was recorded for ZnO originated from the ultrasonically synthesized precursor. Enhancement of removal efficiency that reached 98% was attained in only a period of 8 min. Its recycling test showed that it can be reused without structural changes over four cycling experiments.     

Optimization of extraction and quantification technique for phenolics content of garlic (Allium sativum): An application for comparative phytochemical evaluation based on cultivar origin

A range of conventional, i.e. maceration, percolation, ultrasonic assisted, Soxhlet and Soxtec extraction (STE), to advanced extraction techniques of accelerated solvent extraction (ASE) was utilized for the first time in order to optimize the extract yield and recovery of phenolics—gallic acid (GA), rutin (RT) and quercetin (QT)—quantified via ultra-high performance liquid chromatography with diode array detector (UHPLC–DAD). The effect of solvents (n-hexane, dichloromethane and methanol) and temperature (60, 80 and 100°C) upon extraction yield, phenolic content and antioxidant activity (DPPH, ABTS and DPPH) was studied, and the method was validated in commercial food samples from Saudi Arabia, China and India. A high extract yield with percentage recovery was observed for STE (1221.10 mg/5 g; 24.42%) and ASE techniques (91.50 mg/1 g; 9.15%) in methanol at 100°C. UHPLC–DAD showed retention times (min) of 0.67, 1.93 and 1.90 for GA, RT and QT, respectively in the shortest runtime of 3 min. The yield for phenolics was higher for STE/ASE (ppm): 15.27/15.29 (GA), 85.24/37.56 (RT) and 52.20/33.40 (QT), respectively. In terms of antioxidant activities, low IC₅₀ values (μg/ml) of 1.09/1.18 (DPPH), 2.11/5.32 (ABTS) and 4.35/7.88 (phenazine methosulfate–nicotinamide adenine dinucleotide) were observed for STE and ASE, respectively. Multivariate analysis for STE showed a significant (P = 0.000) correlation for extraction type vs. extract yield and phenolics content; however, there was no significance for antioxidant activities vs. extraction type. ASE showed a positive correlation for solvent vs. extraction yield, phenolics and antioxidant activity; however, there was no correlation for extraction yield and DPPH activity. Principal component analysis for STE showed a major variability (52.02%) for extraction yield and phenolics in PC1 followed by PC2 (38.30%) for antioxidant activities. For ASE, PC1 (48.68%) showed a positive correlation for solvent vs. extraction yield and phenolics while PC2 (33.12%) showed a positive correlation for temperature and antioxidant activities. STE and ASE were the optimized extraction techniques for the garlic food sample while a significant effect of solvent and temperature was observed upon extraction yield, phenolics and antioxidant activity.