Search

Hyperparameter tuning is crucial in the development of machine learning models. This study introduces the nonlinear shrinking factor and the Cauchy mutation mechanism to improve the Dujiangyan Irrigation System Optimization (DISO), proposing the improved Dujiangyan Irrigation System Optimization algorithm (IDISO) for hyperparameter tuning in machine learning. The optimization capabilities and convergence performance of IDISO were validated on 87 CEC2017 benchmark functions of varying dimensions and nine real-world engineering problems, demonstrating that it significantly outperforms DISO in terms of convergence speed and accuracy, and ranks first in overall performance among the seventeen advanced metaheuristic algorithms being compared. To construct a robust and generalizable prediction model for hydrochar element characteristics, this study utilized IDISO and DISO algorithms to fine-tune the parameters of the XGBoost model. The experimental results show that the IDISO-XGBoost model achieved an average prediction performance of 0.95, which represents a 4% improvement over the DISO-XGBoost model. These results indicate that the IDISO algorithm has significant potential and value in practical applications. [ABSTRACT FROM AUTHOR]

News media can influence citizens' health beliefs about COVID-19 and eventually their vaccination intention. However, existing literature has rarely investigated how such effect is contingent upon a country-level factor: press freedom. Situated in the Health Belief Model, this study draws upon a multi-national survey (N = 3,599), involving 10 major cities in Asia to address the research gap. Results showed that news exposure has a positive effect on personal health beliefs on COVID-19, affecting their vaccination intention. More interestingly, the relationship between news exposure and personal health beliefs about COVID-19 was negatively moderated by level of press freedom – that is, the relationship between news exposure and personal health beliefs is stronger in cities that belong to countries with low levels of press freedom. [ABSTRACT FROM AUTHOR]

In this letter, we investigate the trade-off between parasitic capacitance and parasitic resistance of lattice-matched InP high-electron mobility transistors (HEMTs) with a double-recessed gate process. Lg = 80 nm HEMTs with and without SiO2 hardmask are used to adjust parasitic capacitance and parasitic resistance. And their DC and RF responses are measured and analyzed, such as the maximum transconductance (gm_max), ON-resistance (RON), current-gain cutoff frequency (fT), and maximum oscillation frequency (fmax). In order to obtain a high fmax, we adopt the technology of gate recess closer to the source electrode. The device with SiO2 hardmask, which has relatively smaller parasitic resistance, exhibits excellent characteristics of RON = 0.54 Ω·mm, gm_max = 1.087 S/mm, fT = 304 GHz, and fmax = 524 GHz. The device without the SiO2 hardmask achieved smaller drain output conductance (gds) and the ratio of gate-to-drain (Cgd) to gate-to-source (Cgs), resulting in a larger fmax of 690 GHz without significantly lowering fT (285 GHz). The HEMTs are providing more options for circuit processing. [ABSTRACT FROM AUTHOR]

A physical insight into the capture and emission behavior of interface/oxide states in a GaN-based metal-oxide-semiconductor (MOS) structure is of great importance to understanding the threshold voltage (VTH) instability in GaN power transistors. A time-dependent VTH shift in Ni/Al2O3/AlGaN/GaN MOS-HFETs (heterojunction field-effect transistors) and a distribution of Al2O3/III-nitride interface states (Dit) were successfully characterized by constant-capacitance deep level transient spectroscopy. It is found that in situ remote plasma pretreatments in plasma-enhanced atomic-layer-deposition could suppress Dit (EC-ET > 0.4 eV) down to below 1.3 × 1012 cm−2 eV−1. Under high applied gate bias (e.g., VG > 8 V), tunnel filling of oxide states in the Al2O3 dielectric comes into play, contributing to remarkable VTH instability in the MOS-HFETs. The tunnel distance between the 2D Electron Gas (2DEG) channel and oxide states ET,ox in the Al2O3 dielectric decreases from 3.75 to 0.82 nm as VG increases from 2 to 8 V. A further increase of VG to 11 V makes the Fermi level approach ET,ox (EC − ET ∼ 1.62 eV), which may enable direct filling. High electric field induced tunnel filling of gate oxide states could be an assignable cause for VTH instability in normally-OFF III-nitride MOS-HFETs. [ABSTRACT FROM AUTHOR]

Following the COVID-19's outbreak in China, wildlife-related issues such as wildlife management and conservation made headlines around the world due to the potential zoonotic nature of the newly discovered coronavirus. In our study, we examined the dynamic interaction of the news agenda and public agenda concerning wildlife-related issues on social media after COVID-19's outbreak. Using big data analytics, we automatically extracted the agendas' attributes and networks from 110,549 social media posts made from January 1 to April 8, 2020, and investigated the effect of second- and third-level dynamic agenda setting in a time series analysis. Our findings suggest that the agenda-setting effect of wildlife-related issues on social media was not a single-step, unidirectional action but a reciprocal, dynamic interaction constantly constructed by news outlets and the general public. [ABSTRACT FROM AUTHOR]

To identify the psychosocial determinants of individuals' intention to engage in collective actions against cancer, we extended and tested the risk perception attitude (RPA) framework at the level of social perceptions. The results of a large online survey of Hong Kong citizens (N = 1,005) revealed that perceived societal risk and perceived collective efficacy directly and jointly influenced respondents' intention to engage in collective actions against cancer, namely donating to cancer charities, volunteering at cancer-prevention organizations, and supporting public policies for cancer prevention. However, the interaction between perceived societal risk and perceived collective efficacy occurred in a direction opposite to the direction in the initial RPA framework. As suggested by the framework, we also categorized individuals into four attitudinal groups based on their perceptions of societal-level risk as well as efficacy and compared their demographic and psychological characteristics. Among the findings, the four groups significantly differed in their perceptions of individual-level risk as well as efficacy, in their family cancer history, and in their intentions to engage in individual-level behaviors to prevent cancer. Altogether, our findings contribute to the literature by extending the RPA framework to individuals' societal-level perceptions and by providing evidence that the framework can benefit the development of health communication campaigns to promote engagement in collective actions to support cancer prevention. [ABSTRACT FROM AUTHOR]