Initial Bias Research Articles

Dependence of Positive Bias Stress Instability on Threshold Voltage and Its Origin in Solution-Processed Aluminum-Doped Indium Oxide Thin-Film Transistors.

The initial electrical characteristics and bias stabilities of thin-film transistors (TFTs) are vital factors regarding the practical use of electronic devices. In this study, the dependence of positive bias stress (PBS) instability on an initial threshold voltage (VTH) and its origin were analyzed by understanding the roles of slow and fast traps in solution-processed oxide TFTs. To control the initial VTH of oxide TFTs, the indium oxide (InOx) semiconductor was doped with aluminum (Al), which functioned as a carrier suppressor. The concentration of oxygen vacancies decreased as the Al doping concentration increased, causing a positive VTH shift in the InOx TFTs. The VTH shift (∆VTH) caused by PBS increased exponentially when VTH was increased, and a distinct tendency was observed as the gate bias stress increased due to a high vertical electric field in the oxide dielectric. In addition, the recovery behavior was analyzed to reveal the influence of fast and slow traps on ∆VTH by PBS. Results revealed that the effect of the slow trap increased as the VTH moved in the positive direction; this occured because the main electron trap location moved away from the interface as the Fermi level approached the conduction band minimum. Understanding the correlation between VTH and PBS instability can contribute to optimizing the fabrication of oxide TFT-based circuits for electronic applications.

A Theory for Neutron Star and Black Hole Kicks and Induced Spins

Using 20 long-term 3D core-collapse supernova simulations, we find that lower compactness progenitors that explode quasi-spherically due to the short delay to explosion experience smaller neutron star recoil kicks in the ∼100−200 km s−1 range, while higher compactness progenitors that explode later and more aspherically leave neutron stars with kicks in the ∼300−1000 km s−1 range. In addition, we find that these two classes are correlated with the gravitational mass of the neutron star. This correlation suggests that the survival of binary neutron star systems may in part be due to their lower kick speeds. We also find a correlation between the kick and both the mass dipole of the ejecta and the explosion energy. Furthermore, one channel of black hole birth leaves masses of ∼10 M ⊙, is not accompanied by a neutrino-driven explosion, and experiences small kicks. A second channel is through a vigorous explosion that leaves behind a black hole with a mass of ∼3.0 M ⊙ kicked to high speeds. We find that the induced spins of nascent neutron stars range from seconds to ∼10 ms, but do not yet see a significant spin/kick correlation for pulsars. We suggest that if an initial spin biases the explosion direction, a spin/kick correlation would be a common byproduct of the neutrino mechanism of core-collapse supernovae. Finally, the induced spin in explosive black hole formation is likely large and in the collapsar range. This new 3D model suite provides a greatly expanded perspective and appears to explain some observed pulsar properties by default.

Turbofan Engine Health Prediction Model Based on ESO-BP Neural Network

Predicting the health index of turbofan engines is critical in reducing downtime and ensuring aircraft safety. This study introduces the elite snake optimizer-back propagation (ESO-BP) model to address the challenges of low accuracy and poor stability in predicting the health index of turbofan engines through neural networks. Firstly, the snake optimizer (SO) was improved into the elite snake optimizer (ESO) through an elite-guided strategy and a reverse learning mechanism. The performance improvement was validated using benchmark functions. Additionally, feature importance was introduced as a feature selection method. Finally, the optimization results of the ESO were employed to set the initial weights and biases of the BP neural network, preventing convergence to local optima. The prediction performance of the ESO-BP model was validated using the C-MAPSS datasets. The ESO-BP model was compared with the CNN, RNN, LSTM, baseline BP, and unimproved SO-BP models. The results demonstrated that the ESO-BP model has a superior accuracy with an impressive R-squared (R2) value of 0.931 and a root mean square error (RMSE) of 0.060 on the FD001 sub-dataset. Furthermore, the ESO-BP model exhibited lower standard deviations of evaluation metrics on 100 trials. According to the study, ESO-BP demonstrated a greater prediction accuracy and stability when compared to commonly used models such as CNN, RNN, LSTM, and BP.

Optimal estimation of MC parameter in SAE J2601 hydrogen refuelling protocol based on modified formula and artificial neural networks

The MC method was officially released in SAE J2601 protocol in 2016 to refuel light-duty hydrogen fuel cell vehicles quickly and safely. For the non-communication MC method, the pressure target for controlling the end of filling is calculated based on the formula of the final hydrogen temperature, which contains a key parameter, MC. SAE has proposed two formulas to estimate the MC parameter. This article proposes a modified formula and artificial neural networks (ANN) model for MC parameter. To generate simulated data for fitting the modified formula and training the ANN model, a hydrogen filling model (0D1D) with a zero-dimensional gas and a one-dimensional tank wall is established and thoroughly verified. In establishing the ANN model, Latin hypercube sampling (LHS) is used to sample independent variables randomly, and genetic algorithms (GA) are used to optimize the initial weights and biases of the ANN. The Sobol sensitivity analysis is used to determine the importance of various initial conditions on the MC parameter. Research shows that the final fueling time and initial pressure have a greater impact on the MC parameter, while the ambient and precooling temperatures have little impact. The modified formula of the MC parameter improves R2 from the original 0.76073 to 0.98132 when fitting the MC simulated data. In situations of short filling time, the modified formula and ANN models of the MC parameter reduce the relative error between the 0D1D model by 13.8 % − 58.8 % and 16.6 % − 82.3 %, respectively, thus improving the calculation accuracy of the final hydrogen temperature and ending pressure target. Moreover, the ANN model can predict the MC parameter during the whole fueling process, while the MC formula utilizes the fixed value when the fueling time is less than 30 s. This work has positive significance for improving refuelling protocol.

Swin transformer based transfer learning model for predicting porous media permeability from 2D images

Soil and rock, as typical porous media, widely exist in natural slopes and landslides and underground reservoirs. Accurate predicting the permeability of porous media is crucial in preventing geological disasters and extracting underground energy. The conventional methods for obtaining the porous media permeability involve experiments or numerical simulations, both of which are usually time-consuming. In recent years, a new approach to predicting the porous media permeability has emerged, utilizing porous media images and leveraging deep learning algorithms. However, previous image-based permeability prediction frameworks have been dominated by convolutional neural networks (CNNs). CNNs require extensive data and intricate network architectures to achieve reasonably accurate predictive performance. State-of-the-art neural networks and methods are employed to overcome the shortcomings of CNNs. To initiate this research, we generated 1200 two-dimensional porous media images. The permeability labels for the target porous media were acquired through computational fluid dynamics. Subsequently, an end-to-end permeability prediction surrogate model was established based on the Swin Transformer architecture. The proposed model falls under the category of transfer learning models, where pre-trained weights and bias from other datasets serve as the initial weights and bias for our dataset. Our model has achieved State-of-the-art performance, outperforming CNN by a significant margin on small-size datasets. Even with a small dataset size as 200, the proposed model still achieves an R2 score of 0.9717 on the testing set. Both transfer learning and the Swin Transformer architecture have demonstrated robust predictive capabilities in the context of predicting permeability in small-size dataset porous media based on computer vision.

Frequency response of the non-polarization-maintaining depolarized sagnac interferometer based on maximum fringe contrast

The Sagnac Interferometer offers a promising solution to the speckle problem in interferometric detection due to its insensitivity to environmental low-frequency interference, making it well-suited for applications such as laser ultrasonic detection. Control of the polarization state of light is imperative for achieving high sensitivity in optical fiber interferometer and ensuring optimal quiescent operating conditions. This study employed the coherence matrix method to analyze the theory of optimal polarization control in a non-polarization-maintaining depolarized Sagnac Interferometer, encompassing the deduction of the stable π/2 initial phase bias and the maximum fringe contrast of 50% in theory. In experimental observations during PZT simulated large amplitude ultrasound experiments, the fringe contrast reached 48.4%. Furthermore, the system's amplitude-frequency response was evaluated using an electro-optics phase modulator, demonstrating its capability to detect high-frequency signals within 100 MHz These findings provide essential insights into optimizing the Sagnac Interferometer's performance for laser ultrasonic detection and its potential applications in high-frequency signal detection and ultrasonic amplitude measurement.

Improving extreme learning machine model using deep learning feature extraction and grey wolf optimizer: Application to image classification

The Extreme Learning Machine (ELM) is a highly efficient model for real-time network retraining due to its fast learning speed, unlike traditional machine learning methods. However, the performance of ELM can be negatively impacted by the random initialization of weights and biases. Moreover, poor input feature quality can further degrade performance, particularly with complex visual data. To overcome these issues, this paper proposes optimizing the input features as well as the initial weights and biases. We combine both Convolutional Neural Network (CNN) and Convolutional AutoEncoder (CAE) extracted features to optimize the quality of the input features. And we use our hybrid Grey Wolf Optimizer-Multi-Verse Optimizer (GWO-MVO) metaheuristic for initializing weights and biases by applying four fitness functions based on: the norm of the output weights, the error rate on the training set, and the error rate on the validation set. Our method is evaluated on image classification tasks using two benchmark datasets: CIFAR-10 and CIFAR-100. Since image quality may vary in real-world applications, we trained and tested our models on the dataset’s original and noisy versions. The results demonstrate that our method provides a robust and efficient alternative for image classification tasks, offering improved accuracy and reduced overfitting.

Supplemental Material for The Roles of Social Information, Asocial Information, and Initial Bias in Nest-Building Decisions

Supplemental Material for The Roles of Social Information, Asocial Information, and Initial Bias in Nest-Building Decisions

Nudging human drivers via implicit communication by automated vehicles: Empirical evidence and computational cognitive modeling

Understanding behavior of human drivers in interactions with automated vehicles (AV) can aid the development of future AVs. Existing investigations of such behavior have predominantly focused on situations in which an AV a priori needs to take action because the human has the right of way. However, future AVs might need to proactively manage interactions even if they have the right of way over humans, e.g., a human driver taking a left turn in front of the approaching AV. Yet it remains unclear how AVs could behave in such interactions and how humans would react to them. To address this issue, here we investigated behavior of human drivers (N = 19) when interacting with an oncoming AV during unprotected left turns in a driving simulator experiment. We measured the outcomes (Go or Stay) and timing of participants’ decisions when interacting with an AV which performed subtle longitudinal nudging maneuvers, e.g. briefly decelerating and then accelerating back to its original speed. We found that participants’ behavior was sensitive to deceleration nudges but not acceleration nudges. We compared the obtained data to predictions of several variants of a drift-diffusion model of human decision making. The most parsimonious model that captured the data hypothesized noisy integration of dynamic information on time-to-arrival and distance to a fixed decision boundary, with an initial accumulation bias towards the Go decision. Our model not only accounts for the observed behavior but can also flexibly generate predictions of human responses to arbitrary longitudinal AV maneuvers, and can be used for both informing future studies of human behavior and incorporating insights from such studies into computational frameworks for AV interaction planning.

Regulation of craving training to support healthy food choices under stress: A randomized control trial employing the hierarchical drift-diffusion model.

Stress increases the likelihood of consuming unhealthy food in some individuals. Previous research has demonstrated that the Regulation of Craving - Training (ROC-T) intervention can reduce unhealthy food intake. However, its effectiveness under stress and the underlying mechanism remained uncertain. This study aimed to assess the efficacy of the ROC-T intervention in improving healthy food choices and to explore the intervention mechanism through computational modeling employing the hierarchical drift-diffusion model (HDDM). This study adopted a 2 (ROC-T intervention vs. control)*2 (stress vs. no-stress) between-subject experimental design. A total of 118 employees (72 women, Mage=28.74) participated in the online experiment. Results show that the ROC-T intervention increases healthy food choices under stress and no-stress conditions. The HDDM results reveal a significant two-way interaction for non-decision time (Bayes factor, BF = 32.722) and initial bias (BF = 27.350). Specifically, in the no-stress condition, the ROC-T intervention resulted in lower non-decision time and higher initial bias compared with the control group. The findings validated the negative impact of stress on healthy food choices, and that the ROC-T intervention promotes healthy food choices both under stress and no-stress conditions.

Heavy ion single event effect in double-trench SiC metal–oxide–semiconductor field-effect transistors

In this paper, experiments on 208 MeV Ge ion irradiation with different source-drain bias voltages are carried out for the double-trench SiC metal–oxide–semiconductor field-effect transistors, and the physical mechanism of the single event effect is analyzed. The experimental results show that the drain leakage current of the device increases more obviously with the increase of the initial bias voltage during irradiation. When the bias voltage is 400 V during irradiation, the device has a single event burned at a fluence of 9×10<sup>4</sup> ion/cm<sup>2</sup>, and when the bias voltage is 500 V, the device has a single event burned at a fluence of 3×10<sup>4</sup> ion/cm<sup>2</sup>, so when the LET value is 37.3 MeV·cm<sup>2</sup>/mg, the SEB threshold of DUT does not exceed 400 V, which is lower than 34 % of the rated operational voltage. The post gate-characteristics test results show that the leakage current of the device with a bias voltage of 100 V does not change significantly during irradiation. When the bias voltage is 200 V, the gate leakage and the drain leakage of the device both increase, so do they when the bias voltage is 300 V, which is positively related to bias voltage. In order to further analyze the single particle effect mechanism of device, the simulation is conducted by using TCAD tool. The simulation results show that at low bias voltage, the heavy ion incident device generates electron-hole pairs, the electrons are quickly swept out, and the holes accumulate at the gate oxygen corner under the effect of the electric field, which combines the source-drain bias voltage, leading to the formation of leakage current channels in the gate oxygen layer. The simulation results also show that at high bias voltage, the electrons generated by the incident heavy ion move towards the junction of the N<sup>–</sup> drift layer and the N<sup>+</sup> substrate under the effect of the electric field, which further increases the electric field strength and causes significant impact ionization. The local high current density generated by the impact ionization and the load large electric field causes the lattice temperature to exceed the melting point of silicon carbide, causing single event burnout. This work provides a reference and support for studying the radiation effect mechanism and putting silicon carbide power devices into aerospace applications.

Replication elongates short DNA, reduces sequence bias and develops trimer structure.

The origin of molecular evolution required the replication of short oligonucleotides to form longer polymers. Prebiotically plausible oligonucleotide pools tend to contain more of some nucleobases than others. It has been unclear whether this initial bias persists and how it affects replication. To investigate this, we examined the evolution of 12-mer biased short DNA pools using an enzymatic model system. This allowed us to study the long timescales involved in evolution, since it is not yet possible with currently investigated prebiotic replication chemistries. Our analysis using next-generation sequencing from different time points revealed that the initial nucleotide bias of the pool disappeared in the elongated pool after isothermal replication. In contrast, the nucleotide composition at each position in the elongated sequences remained biased and varied with both position and initial bias. Furthermore, we observed the emergence of highly periodic dimer and trimer motifs in the rapidly elongated sequences. This shift in nucleotide composition and the emergence of structure through templated replication could help explain how biased prebiotic pools could undergo molecular evolution and lead to complex functional nucleic acids.

Innovative fault diagnosis and early warning method based on multifeature fusion model for electric vehicles

Fault diagnostics and early warning are crucial to the safe operation of lithium-ion batteries, and despite partial progress, it is still extremely difficult to solve the problem in a high-dimensional parameter space with complex failure causes. In this article, an innovative fault diagnosis and early warning method based on multifeature fusion model is designed for quantitative and qualitative comprehensive analysis and evaluation of the battery operating state information and the complex internal safety evolution trajectory. Firstly, to form a multi-dimensional fault feature matrix, a real vehicle dataset consisting of 400 vehicles is constructed using typical vehicle fault data from the cloud platform. Then, the original data is noise-reduced and pre-processed by applying initial bias correction and anomalous signal identification methods. Moreover, based on the equivalent circuit model and data-driven algorithm, 13 features including internal short circuit internal resistance, inconsistency accumulation risk, and principal element extraction index are designed from three dimensions: threshold, statistics and model. Finally, the multiple features are combined with random forest (RF) algorithm for hybrid fusion decision to achieve better generalization ability and suppress the overfitting risk of the algorithm. The results show that the proposed feature extraction and fusion decision methods can identify abnormal states and hazard levels in a timely and accurate manner, and this RF-based classification, warning and evaluation framework shows the promise of machine learning algorithms for interpretable early warning studies of battery failures, which can use more modal features to assess the hazard levels of actual operating battery systems in end-cloud collaboration scenarios.

Reference Motion Quality and Design Choices for Bipedal Walking with PPO

This paper investigates the influence of reference motion quality and other design choices on the performance of deep reinforcement learning for bipedal walking with Proximate Policy Optimization (PPO). We use parametrized Cartesian quintic splines to generate reference actions for an omnidirectional walk policy. By using parameter sets with different qualities, we show that the performance of the trained policy correlates to the quality of the reference motion. We also show that a policy in Cartesian space performs superior to a joint-space-based one if an advantageous representation of orientation is chosen. Additionally, we show that using an initial bias for the policy speeds up the training and leads to higher performances for policies using position control. We also show that we can achieve a stable omnidirectional walk on a wide variety of simulated humanoid robots.

Initialization shock in the ocean circulation reduces skill in decadal predictions of the North Atlantic subpolar gyre

Due to large northward heat transport, the Atlantic meridional overturning circulation (AMOC) strongly affects the climate of various regions. Its internal variability has been shown to be predictable decades ahead within climate models, providing the hope that synchronizing ocean circulation with observations can improve decadal predictions, notably of the North Atlantic subpolar gyre (SPG). Climate predictions require a starting point which is a reconstruction of the past climate. This is usually performed with data assimilation methods that blend available observations and climate model states together. There is no unique method to derive the initial conditions. Moreover, this can be performed using full-field observations or their anomalies superimposed on the model's climatology to avoid strong drifts in predictions. How critical ocean circulation drifts are for prediction skill has not been assessed yet. We analyze this possible connection using the dataset of 12 decadal prediction systems from the World Meteorological Organization Lead Centre for Annual-to-Decadal Climate Prediction. We find a variety of initial AMOC errors within the predictions related to a dynamically imbalanced ocean states leading to strongly displaced or multiple maxima in the overturning structures. This likely results in a blend of what is known as model drift and initial shock. We identify that the AMOC initialization influences the quality of the SPG predictions. When predictions show a large initial error in their AMOC, they usually have low skill for predicting internal variability of the SPG for a time horizon of 6-10 years. Full-field initialized predictions with low AMOC drift show better SPG skill than those with a large AMOC drift. Nevertheless, while the anomaly-initialized predictions do not experience large drifts, they show low SPG skill when skill also present in historical runs is removed using a residual correlation metric. Thus, reducing initial shock and model biases for the ocean circulation in prediction systems might help to improve their prediction for the SPG beyond 5 years. Climate predictions could also benefit from quality-check procedure for assimilation/initialization because currently the research groups only reveal the problems in initialization once the set of predictions has been completed, which is an expensive effort.

Missing out by pursuing rewarding outcomes: Why initial biases can lead to persistent suboptimal choices.

Missing out by pursuing rewarding outcomes: Why initial biases can lead to persistent suboptimal choices.

Evaluation of future climatology and its uncertainty under SSP scenarios based on a bias processing procedure: A case study of the Lancang-Mekong River Basin

The Lancang-Mekong River is the most important international river in Southeast Asia, and its water resources are invaluable for the survival of the inhabitants of the basin, while the climatic conditions have a significant impact on the production and livelihoods of the riparian countries. In this study, future climate change in the Lancang-Mekong River Basin (LMRB) was projected based on precipitation and temperature from CMIP6. First, a bias processing procedure was established to correct the initial bias of climate models using the Daily Bias Correction (DBC) method, and the performance of equal weighted averaging (EW) and Bayesian mode averaging (BMA) of multi-model ensembles was compared. The characteristics of future precipitation and temperature changes were projected in terms of mean climatology, seasonal variation and extreme events, respectively, from which the future trends of the four weather extremes (dry-cold, dry-hot, wet-cold and wet-hot) were projected. Evaluation based on mean climatology, daily frequency distribution, temporal change and its bias showed that the climate bias correction procedure has good performance. It was found that future precipitation in the LMRB will increase by 50–120 mm in the near future (2031–2050) and 150–400 mm in the far future (2061–2090), with higher increases in the upper Mekong River Basin (MRB), Lower MRB and Mekong Delta than in the Lancang River Basin (LRB) and middle MRB. Future increase in precipitation will occur mainly from June to October, while April and May will see a decrease by 10–30 mm, and the number of heavy precipitation days (R25, daily precipitation exceeding 25 mm) will increase by about 5 days in the eastern part of the lower and middle MRB. The mean temperature will increase by 1–3 °C in most parts of the LMRB, with the LRB experiencing an increase of about 3 °C higher than in the MRB. Future maximum and minimum daily temperatures will increase by 1–2 °C in the near future and by 2–5 °C in the far future. Dry-hot days will increase by about 20 days in the lower and middle MRB, and wet-hot days will increase significantly in the MRB, by 30–80 days in the near future and by 50–100 days in the far future.

A robust vector tracking loop structure based on potential bias analysis

This paper proposes a robust vector tracking loop structure based on potential bias analysis. The influence of four kinds of biases on the existing two implementations of Vector Tracking Loops (VTLs) is illustrated by theoretical analysis and numerical simulations, and the following findings are obtained. Firstly, the initial user state bias leads to steady navigation solution bias in the relative VTL, while new measurements can eliminate it in the absolute VTL. Secondly, the initial code phase bias is transferred to the following navigation solutions in the relative VTL, while new measurements can eliminate it in the absolute VTL. Thirdly, the user state bias induced by erroneous navigation solution of VTLs can be eliminated by both of the two VTLs. Fourthly, the multipath/NLOS likely affects the two VTLs, and the induced tracking bias in the duration of the multipath/NLOS would decrease the performance of VTLs. Based on the above analysis, a robust VTL structure is proposed, where the absolute VTL is selected for its robustness to the two kinds of initialization biases; meanwhile, the instant bias detection and correction method is used to improve the performance of VTLs in the duration of the multipath/NLOS. Numerical simulations and experimental results verify the effectiveness of the proposed robust VTL structure.

Optimized deep extreme learning machine for traffic prediction and autonomous vehicle lane change decision-making

Deep learning occupies a prominent position in the field of machine learning, where researchers have introduced autoencoder into extreme learning machine (ELM) to establish deep extreme learning machine (DELM), with the aim of enabling autonomous feature learning from data. This approach has found widespread applications in various domains, including finance and electrochemistry. However, the network accuracy of DELM is sensitive to the initial weighting and bias configurations, which can be negatively impacted by random initialization. To mitigate the adverse effects of such random initialization on network performance, this study employs two swarm intelligence optimization algorithms, particle swarm optimization (PSO) and harris hawks optimization (HHO), to optimize the input weights and biases of DELM. In addition, we apply the variational mode decomposition (VMD) technique to decompose the original data, enabling a deeper exploration of latent information within the data. Subsequently, two integrated models are constructed: VMD-HHO-DELM and VMD-PSO-DELM. We apply these models to address road traffic condition prediction and autonomous lane change decision-making problems, with a specific focus on the transportation domain. Through extensive experimentation on two real-world datasets, we observe that deep learning models based on swarm intelligence optimization algorithms exhibit excellent performance across various scenarios. In the context of road traffic condition prediction, our experiments reveal superior results under different time intervals and traffic volumes compared to traditional models, all while maintaining higher operational efficiency. For the autonomous lane change decision experiment, we establish a driver dissatisfaction model that aids in formulating more reasonable lane change decisions, thus reducing the frequency of abrupt lane changes.

Selection bias in multi-technology auctions: How to quantify and assess efficiency implications in renewable energy auctions

We develop a concept to identify and quantitatively assess technology selection bias in multi-technology renewable energy auctions. We show that simple price rules are insufficient to efficiently select winning projects in multi-technology tenders when they incorporate individual costs of producers only, and exclude system effects, market benefits and external costs. With our concept of quantifying unit social value, all relevant external elements can be incorporated in the evaluation. The introduction of an objective measure (as difference of differences between unit social value and auction bid prices) allows the quantitative assessment of systematic selection bias between technologies under different remuneration designs. We illustrate our concept by applying it to generic European renewable energy technologies (wind and solar energy) and major applied remuneration types (contract-for-difference and fixed premium schemes). We also show that are concept is applicable for real auction data, by presenting a case study about Italy. The main conclusion of the study is that selection by price only, can constitute a systematic bias for all investigated remuneration designs. Both schemes are in our case biased toward a given technology. The internalisation of external cost may not necessarily lead to better social outcomes in the selection of auction winners, as it can be overshadowed by the initial bias of the chosen remuneration design. We conclude that considerate design, including the potential differentiation of rules, e.g., through introduction of a technology bonus, to minimise selection bias, and careful monitoring of the competitive situation between participating technologies in a multi-technology auction are required to ensure efficient selection.