Evolving Least Squares Support Vector Machines for Stock Market Trend Mining

In this paper, an evolving least squares support vector machine (LSSVM) learning paradigm with a mixed kernel is proposed to explore stock market trends. In the proposed learning paradigm, a genetic algorithm (GA), one of the most popular evolutionary algorithms (EAs), is first used to select input features for LSSVM learning, i.e., evolution of input features. Then, another GA is used for parameters optimization of LSSVM, i.e., evolution of algorithmic parameters. Finally, the evolving LSSVM learning paradigm with best feature subset, optimal parameters, and a mixed kernel is used to predict stock market movement direction in terms of historical data series. For illustration and evaluation purposes, three important stock indices, S&P 500 Index, Dow Jones Industrial Average (DJIA) Index, and New York Stock Exchange (NYSE) Index, are used as testing targets. Experimental results obtained reveal that the proposed evolving LSSVM can produce some forecasting models that are easier to be interpreted by using a small number of predictive features and are more efficient than other parameter optimization methods. Furthermore, the produced forecasting model can significantly outperform other forecasting models listed in this paper in terms of the hit ratio. These findings imply that the proposed evolving LSSVM learning paradigm can be used as a promising approach to stock market tendency exploration.     

CoBTx-Net: A model for business collaboration reliability verification

Collaborative business process can become unreliable when business partners collaborate in a peer- based fashion without central control. Therefore, reliability checking becomes an important issue that needs to be dealt with for any generic solution in managing business collaboration. In this paper, we propose a novel Choreographical Business Transaction Net (CoBTx-Net) to model collaborative business process and to manage the collaboration by individual participants. Furthermore three reliability properties named Time-embedded dead marking freeness, Inter-organizational dead marking freeness, and Collaborative soundness are defined based on CoBTx-Net to verify (1) the violation of time constraint, (2) collaborative logic conflicts, and (3) the improper termination from individual organizations.

Credit scoring using support vector machines with direct search for parameters selection

Support vector machines (SVM) is an effective tool for building good credit scoring models. However, the performance of the model depends on its parameters’ setting. In this study, we use direct search method to optimize the SVM-based credit scoring model and compare it with other three parameters optimization methods, such as grid search, method based on design of experiment (DOE) and genetic algorithm (GA). Two real-world credit datasets are selected to demonstrate the effectiveness and feasibility of the method. The results show that the direct search method can find the effective model with high classification accuracy and good robustness and keep less dependency on the initial search space or point setting.     

Maximum Penalized Likelihood Kernel Regression for Fast Adaptation

This paper proposes a nonlinear generalization of the popular maximum-likelihood linear regression (MLLR) adaptation algorithm using kernel methods. The proposed method, called maximum penalized likelihood kernel regression adaptation (MPLKR), applies kernel regression with appropriate regularization to determine the affine model transform in a kernel-induced high-dimensional feature space. Although this is not the first attempt of applying kernel methods to conventional linear adaptation algorithms, unlike most of other kernelized adaptation methods such as kernel eigenvoice or kernel eigen-MLLR, MPLKR has the advantage that it is a convex optimization and its solution is always guaranteed to be globally optimal. In fact, the adapted Gaussian means can be obtained analytically by simply solving a system of linear equations. From the Bayesian perspective, MPLKR can also be considered as the kernel version of maximum a posteriori linear regression (MAPLR) adaptation. Supervised and unsupervised speaker adaptation using MPLKR were evaluated on the Resource Management and Wall Street Journal 5K tasks, respectively, achieving a word error rate reduction of 23.6% and 15.5% respectively over the speaker-independently model.     

Use of the fast Fourier transform method for analyzing linear and equispaced Fizeau fringes

The Fourier transform method is applied to analyze the initial phase of linear and equispaced Fizeau fringes. We develop an algorithm for high-precision phase measurement by using the Fourier coefficient that corresponds to the spatial frequency of the Fizeau fringes, and we describe methods for determining the fringe carrier frequency. Errors caused by carrier frequency fluctuation and data truncation are studied theoretically and by computer simulation. To demonstrate the method we apply it to the real-time calibration of a piezoelectric transducer mirror in a Twyman-Green interferometer.     

Phase distribution and associated mechanical property distribution in silicon carbide particle-reinforced aluminium fabricated by liquid metal infiltration

The reaction between silicon carbide and aluminium to form silicon and Al₄C₃ in SiC particle-reinforced aluminium fabricated by liquid aluminium infiltration was most severe near the original interface between liquid aluminium and the SiC preform. This resulted in the highest concentration of Al₄C₃ and the lowest concentrations of silicon and SiC in the part of the composite near this interface. In particular, the silicon concentration was highest in the bottom centre of the composite when infiltration occurred from the top, because silicon diffused toward the surrounding aluminium melt before solidification. These non-uniform phase distributions, as measured by X-ray diffraction and differential scanning calorimetry, did not cause any non-uniform shear strength distribution. However, excessive reaction between SiC and aluminium, as observed for an infiltration (=mould=liquid metal) temperature of 780° C, caused the tensile strength to decrease. In the case where a steel mould was used during infiltration at 780° C, iron-containing precipitates, such as ternary Al-Fe-Si, were observed in the part of the composite within 5 mm from the above-mentioned interface; their formation was related to the silicon out-diffusion in the form of liquid Al-Si; they caused the shear strength to be lower in this part of the composite; larger such precipitates (up to 100 m) were observed in the excess aluminium adjacent to the cast composite. For pure aluminium as the infiltrating metal, the optimum infiltration temperature for the highest tensile strength was 700° C. An infiltration temperature of 670° C resulted in incomplete infiltration, which was more severe when a steel mould rather than a graphite mould was used because of the higher thermal conductivity of the former.     

RADIOIMMUNOTOXICOLOGICAL EFFECT OF ENRICHED URANIUM ON CENTRAL AND PERIPHERAL IMMUNE CELLS AND THE PROTECTIVE ACTION OF IL－1 AND IL－2

ＲＡＤＩＯＩＭＭＵＮＯＴＯＸＩＣＯＬＯＧＩＣＡＬＥＦＦＥＣＴＯＦＥＮＲＩＣＨＥＤＵＲＡＮＩＵＭＯＮＣＥＮＴＲＡＬＡＮＤＰＥＲＩＰＨＥＲＡＬＩＭＭＵＮＥＣＥＬＬＳＡＮＤＴＨＥＰＲＯＴＥＣＴＩＶＥＡＣＴＩＯＮＯＦＩＬ－１ＡＮＤＩＬ－２￥ＺｈｕＳｈｏｕｐｅ...     

The Edge Effects Boosting Hydrogen Evolution Performance of Platinum/Transition Bimetallic Phosphide Hybrid Electrocatalysts

Platinum (Pt) is regarded as a promising electrocatalyst for hydrogen evolution reaction (HER). However, its application in an alkaline medium is limited by the activation energy of water dissociation, diffusion of H⁺, and desorption of H*. Moreover, the formation of effective structures with a low Pt usage amount is still a challenge. Herein, guided by the simulation discovery that the edge effect can boost local electric field (LEF) of the electrocatalysts for faster proton diffusion, platinum nanocrystals on the edge of transition metal phosphide nanosheets are fabricated. The unique heterostructure with ultralow Pt amount delivered an outstanding HER performance in an alkaline medium with a small overpotential of 44.5 mV and excellent stability for 80 h at the current density of −10 mA cm⁻². The mass activity of as-prepared electrocatalyst is 2.77 A mg⁻¹_Pt, which is 15 times higher than that of commercial Pt/C electrocatalysts (0.18 A mg⁻¹_Pt). The density function theory calculation revealed the efficient water dissociation, fast adsorption, and desorption of protons with hybrid structure. The study provides an innovative strategy to design unique nanostructures for boosting HER performances via achieving both synergistic effects from hybrid components and enhanced LEF from the structural edge effect.     

Flexible Transparent Bifunctional Capacitive Sensors with Superior Areal Capacitance and Sensing Capability based on PEDOT:PSS/MXene/Ag Grid Hybrid Electrodes

Flexible transparent supercapacitors (FTSs) have aroused considerable attention. Nonetheless, balancing energy storage capability and transparency remains challenging. Herein, a new type of FTSs with both excellent energy storage and superior transparency is developed based on PEDOT:PSS/MXene/Ag grid ternary hybrid electrodes. The hybrid electrodes can synergistically utilize the high optoelectronic properties of Ag grids, the excellent capacitive performance of MXenes, and the superior chemical stability of PEDOT:PSS, thus, simultaneously demonstrating excellent optoelectronic properties (T: ≈89%, R_s: ≈39 Ω sq⁻¹), high areal specific capacitance, superior mechanical softness, and excellent anti-oxidation capability. Due to the excellent comprehensive performances of the hybrid electrodes, the resulting FTSs exhibit both high optical transparency (≈71% and ≈60%) and large areal specific capacitance (≈3.7 and ≈12 mF cm⁻²) besides superior energy storage capacity (P: 200.93, E: 0.24 µWh cm⁻²). Notably, the FTSs show not only excellent energy storage but also exceptional sensing capability, viable for human activity recognition. This is the first time to achieve FTSs that combine high transparency, excellent energy storage and good sensing all-in-one, which make them stand out from conventional flexible supercapacitors and promising for next-generation smart flexible energy storage devices.     

Bacteria-Responsive Self-Assembly of Antimicrobial Peptide Nanonets for Trap-and-Kill of Antibiotic-Resistant Strains

Bacterial trapping using nanonets is a ubiquitous immune defense mechanism against infectious microbes. These nanonets can entrap microbial cells, effectively arresting their dissemination and rendering them more vulnerable to locally secreted microbicides. Inspired by this evolutionarily conserved anti-infective strategy, a series of 15 to 16 residue-long synthetic β-hairpin peptides is herein constructed with the ability to self-assemble into nanonets in response to the presence of bacteria, enabling spatiotemporal control over microbial killing. Using amyloid-specific K114 assay and confocal microscopy, the membrane components lipoteichoic acid and lipopolysaccharide are shown to play a major role in determining the amyloid-nucleating capacity as triggered by Gram-positive and Gram-negative bacteria respectively. These nanonets displayed both trapping and killing functionalities, hence offering a direct improvement from the trap-only biomimetics in literature. By substituting a single turn residue of the non-amyloidogenic BTT1 peptide, the nanonet-forming BTT1-3A analog is produced with comparable antimicrobial potency. With the same sequence manipulation approach, BTT2-4A analog modified from BTT2 peptide showed improved antimicrobial potency against colistin-resistant clinical isolates. The peptide nanonets also demonstrated robust stability against proteolytic degradation, and promising in vivo efficacy and biosafety profile. Overall, these bacteria-responsive peptide nanonets are promising clinical anti-infective alternatives for circumventing antibiotic resistance.