High Accuracy Of Model Research Articles

Explainable artificial intelligence for fitness prediction of young athletes living in unfavorable environmental conditions

This study is the first attempt to use explainable machine learning (ML) to investigate the factors influencing physical development and the athlete's performance in the Karakalpakstan and Khwarazm regions of the Republic of Uzbekistan. Three physical fitness tests (Dribbling Shuttle, Goal Accuracy, and Yo-Yo intermittent), anthropometrics, and hematology data (33 subject-specific factors) of 60 young soccer players were used in this study. Results reveal that the explainable ML algorithms were effective in prediction with high model accuracy measured from mean absolute error and mean squared error (0.004 and 0.052 for Dribbling Shuttle test with Random Forest model, 41.783 and 5 for Yo-Yo Intermittent Recovery Test-level with XGB model and 0.027 and 0.135 for Goal Accuracy test with XGB model). In contrast, model bias was estimated using mean absolute percentage error (3.6 % for the Dribbling Shuttle test with the Random Forest model, 0.3 % for the Yo-Yo Intermittent Recovery Test with the XGB model and 36.7 % for the Goal Accuracy test with the XGB model). Furthermore, Shapley additive explanations analysis exhibited that the anthropometric parameters such as body mass, mesomorph, and ankle diameter are dominant in physical development. These results confirm the generally accepted concept of the dominant anthropometric profile of early maturing professional soccer players. Additionally, it was understood that the hematology parameters of red blood count and mean corpuscular volume significantly influence physical development. Overall, explainable AI-based methods are more effective in predicting young athletes' physical development and performance. Therefore, the findings can effectively be incorporated with fitness assessment interventions and enhance related regional policy development.

Dynamics Modeling and Parameter Identification for a Coupled-Drive Dual-Arm Nursing Robot

A dual-arm nursing robot can gently lift patients and transfer them between a bed and a wheelchair. With its lightweight design, high load-bearing capacity, and smooth surface, the coupled-drive joint is particularly well suited for these robots. However, the coupled nature of the joint disrupts the direct linear relationship between the input and output torques, posing challenges for dynamic modeling and practical applications. This study investigated the transmission mechanism of this joint and employed the Lagrangian method to construct a dynamic model of its internal dynamics. Building on this foundation, the Newton-Euler method was used to develop a dynamic model for the entire robotic arm. A continuously differentiable friction model was incorporated to reduce the vibrations caused by speed transitions to zero. An experimental method was designed to compensate for gravity, inertia, and modeling errors to identify the parameters of the friction model. This method establishes a mapping relationship between the friction force and motor current. In addition, a Fourier series-based excitation trajectory was developed to facilitate the identification of the dynamic model parameters of the robotic arm. Trajectory tracking experiments were conducted during the experimental validation phase, demonstrating the high accuracy of the dynamic model and the parameter identification method for the robotic arm. This study presents a dynamic modeling and parameter identification method for coupled-drive joint robotic arms, thereby establishing a foundation for motion control in humanoid nursing robots.

Rapid and accurate identification of structural and electronic properties of Xenes under different strains via a first-principles machine-learning approach for energy-electronic devices

Mono-elemental two-dimensional materials (Xenes) are of supreme importance to new-type energy-electronic devices due to their outstanding and unique physical and electronic properties. However, the rapid and accurate identification of the structural and electronic properties of Xenes under various strains remains a challenge, significantly impeding the development of Xene-based energy-electronic devices. Here, we present an effective first-principles machine-learning approach for developing models based on calculated datasets to rapidly and accurately predict the structural and electronic properties of Xenes under different strains. Twelve descriptors with well-defined physical meanings are established. Four different machine-learning algorithms are selected to build models to identify crucial band-gaps. The Decision Tree Regression (DTR) algorithm performs best on both the training and test sets. Tellurene is chosen as the subject of investigation to validate the accuracy of the DTR model. Compared to the DFT computed result (1.230 eV), the DTR model predicted a band-gap of 1.234 eV for Tellurene, demonstrating the model's high reliability and accuracy. Additionally, the valence of the element in Xenes is the most important feature in the DTR model. Our work demonstrates a promising first-principles machine-learning approach, representing a crucial stride towards accelerating the prediction of key-component properties under different strains for Xene-based energy-electronic devices.

A real-time and high-accuracy railway obstacle detection method using lightweight CNN and improved transformer

With the sustained development of railway transportation, the urgent necessity to improve train operation safety makes obstacle detection become the research focus. However, existing railway obstacle detectors still face challenges in balancing detection accuracy and speed during the shunting process. In addition, they are not robust enough in real-world railway environments, especially in complex scenes involving small obstacles. To address these problems, this paper presents a real-time and high-accuracy railway obstacle detection model using lightweight CNN and improved transformer (RH-Net) for detecting railway obstacles efficiently to guarantee traffic safety. First, the Lightweight Feature Extraction Module (LEM) is designed to minimize the model’s computational load while maintaining its feature extraction ability. Then, the Improved Transformer Module (IFM) is developed to boost the model’s ability about stably extract global contextual information. Finally, the Enhanced Multi-Scale Feature Fusion Module (EFM) is proposed to optimize the detection of obstacles with different sizes, especially small objects. In the experiments on railway dataset, RH-Net achieves optimal detection performance on GeForce GTX 1080Ti (96.99% mAP and 135 FPS) and Jetson Xavier NX (97.02% mAP and 43 FPS), which is significantly superior to the existing detection models. Experimental results show that RH-Net has excellent detection ability, which can accurately and efficiently detect obstacles in complicated railway environments. Moreover, the experiments on MS COCO indicate that RH-Net can achieve more satisfactory detection performance than existing state-of-the-art methods. Therefore, the proposed model can be well-applied to more complex real-world scenes for multiple object detection.

Machine learning-based non-destructive terahertz detection of seed quality in peanut

Rapid identification of peanut seed quality is crucial for public health. In this study, we present a terahertz wave imaging system using a convolutional neural network (CNN) machine learning approach. Terahertz waves are capable of penetrating the seed shell to identify the quality of peanuts without causing any damage to the seeds. The specificity of seed quality on terahertz wave images is investigated, and the image characteristics of five different qualities are summarized. Terahertz wave images are digitized and used for training and testing of convolutional neural networks, resulting in a high model accuracy of 98.7% in quality identification. The trained THz-CNNs system can accurately identify standard, mildewed, defective, dried and germinated seeds, with an average detection time of 2.2 s. This process does not require any sample preparation steps such as concentration or culture. Our method swiftly and accurately assesses shelled seed quality non-destructively.

Differentiating loss of consciousness causes through artificial intelligence-enabled decoding of functional connectivity

Differential diagnosis of acute loss of consciousness (LOC) is crucial due to the need for different therapeutic strategies despite similar clinical presentations among etiologies such as nonconvulsive status epilepticus, metabolic encephalopathy, and benzodiazepine intoxication. While altered functional connectivity (FC) plays a pivotal role in the pathophysiology of LOC, there has been a lack of efforts to develop differential diagnosis artificial intelligence (AI) models that feature the distinctive FC change patterns specific to each LOC cause. Three approaches were applied for extracting features for the AI models: three-dimensional FC adjacency matrices, vectorized FC values, and graph theoretical measurements. Deep learning using convolutional neural networks (CNN) and various machine learning algorithms were implemented to compare classification accuracy using electroencephalography (EEG) data with different epoch sizes. The CNN model using FC adjacency matrices achieved the highest accuracy with an AUC of 0.905, with 20-s epoch data being optimal for classifying the different LOC causes. The high accuracy of the CNN model was maintained in a prospective cohort. Key distinguishing features among the LOC causes were found in the delta and theta brain wave bands. This research advances the understanding of LOC's underlying mechanisms and shows promise for enhancing diagnosis and treatment selection. Moreover, the AI models can provide accurate LOC differentiation with a relatively small amount of EEG data in 20-s epochs, which may be clinically useful.

Multi-layer non-hydrostatic free-surface flow model with kinematic seafloor for seismic tsunami generation

Earthquake induced tsunamis pose devastating threat to coastal communities worldwide. Accurate description of tsunami generation by kinematic fault rupture is of importance to investigate earthquake events and evaluate tsunami impacts. The paper develops a multi-layer non-hydrostatic model with a kinematic bottom boundary condition and conducts comprehensive validation to examine the model’s capability in resolving seismic tsunami generation. The two-dimensional governing equations of the multi-layer non-hydrostatic free-surface flow system equipped with kinematic seafloor displacement is first introduced in Cartesian coordinates. A combined finite difference and finite volume scheme is utilized to discretize the governing equations with flow variables arranged on a staggered Arakawa C-grid. Application of pressure correction technique to the discretized formulations yields Poisson-type equations from which the non-hydrostatic pressure is solved for the next time step to complete the temporal integration. Based on existing analytical solutions, four groups of tsunami generation cases considering broad ranges of source parameters, including horizontal scale, rise time, and rupture velocity, are designed to demonstrate performance of the proposed non-hydrostatic model with one-, two-, and three-layers as well as the hydrostatic one. Comparison of 59 generation cases including extreme scenarios indicates the non-hydrostatic model performs better than the hydrostatic model in reproducing the entire waveform and predicting the maximum wave amplitude. High modeling accuracy can be achieved through incorporation of more layers. The proposed multi-layer non-hydrostatic model is a powerful tool for investigating earthquake source mechanisms and evaluating coastal tsunami hazards.

Prediction of Geometric Characteristics of Laser Cladding Layer Based on Least Squares Support Vector Regression and Crested Porcupine Optimization.

The morphology size of laser cladding is a crucial parameter that significantly impacts the quality and performance of the cladding layer. This study proposes a predictive model for the cladding morphology size based on the Least Squares Support Vector Regression (LSSVR) and the Crowned Porcupine Optimization (CPO) algorithm. Specifically, the proposed model takes three key parameters as inputs: laser power, scanning speed, and powder feeding rate, with the width and height of the cladding layer as outputs. To further enhance the predictive accuracy of the LSSVR model, a CPO-based optimization strategy is applied to adjust the penalty factor and kernel parameters. Consequently, the CPO-LSSVR model is established and evaluated against the LSSVR model and the Genetic Algorithm-optimized Backpropagation Neural Network (GA-BP) model in terms of relative error metrics. The experimental results demonstrate that the CPO-LSSVR model can achieve a significantly improved relative error of no more than 2.5%, indicating a substantial enhancement in predictive accuracy compared to other methods and showcasing its superior predictive performance. The high accuracy of the CPO-LSSVR model can effectively guide the selection of laser cladding process parameters and thereby enhance the quality and efficiency of the cladding process.

Enhancing the Performance of Machine Learning and Deep Learning-Based Flood Susceptibility Models by Integrating Grey Wolf Optimizer (GWO) Algorithm

Flooding is a recurrent hazard occurring worldwide, resulting in severe losses. The preparation of a flood susceptibility map is a non-structural approach to flood management before its occurrence. With recent advances in artificial intelligence, achieving a high-accuracy model for flood susceptibility mapping (FSM) is challenging. Therefore, in this study, various artificial intelligence approaches have been utilized to achieve optimal accuracy in flood susceptibility modeling to address this challenge. By incorporating the grey wolf optimizer (GWO) metaheuristic algorithm into various models—including recurrent neural networks (RNNs), support vector regression (SVR), and extreme gradient boosting (XGBoost)—the objective of this modeling is to generate flood susceptibility maps and evaluate the variation in model performance. The tropical Manimala River Basin in India, severely battered by flooding in the past, has been selected as the test site. This modeling utilized 15 conditioning factors such as aspect, enhanced built-up and bareness index (EBBI), slope, elevation, geomorphology, normalized difference water index (NDWI), plan curvature, profile curvature, soil adjusted vegetation index (SAVI), stream density, soil texture, stream power index (SPI), terrain ruggedness index (TRI), land use/land cover (LULC) and topographic wetness index (TWI). Thus, six susceptibility maps are produced by applying the RNN, SVR, XGBoost, RNN-GWO, SVR-GWO, and XGBoost-GWO models. All six models exhibited outstanding (AUC above 0.90) performance, and the performance ranks in the following order: RNN-GWO (AUC: 0.968) > XGBoost-GWO (AUC: 0.961) > SVR-GWO (AUC: 0.960) > RNN (AUC: 0.956) > XGBoost (AUC: 0.953) > SVR (AUC: 0.948). It was discovered that the hybrid GWO optimization algorithm improved the performance of three models. The RNN-GWO-based flood susceptibility map shows that 8.05% of the MRB is very susceptible to floods. The modeling found that the SPI, geomorphology, LULC, stream density, and TWI are the top five influential conditioning factors.

Machine Learning-Aided Analysis of the Rolling and Recrystallization Textures of Pure Iron with Different Cold Reduction Ratios and Cold-Rolling Directions.

We performed a machine learning-aided analysis of the rolling and recrystallization textures in pure iron with different cold reduction ratios and cold-rolling directions. Five types of specimens with different cold reduction ratios and cold-rolling directions were prepared. The effect of two-way cold-rolling on the rolling texture was small at cold reduction ratios different from 60%. The cold reduction ratio in each stage hardly affected the texture evolution during cold-rolling and subsequent short-term annealing. In the case of long-term annealing, although abnormal grain growth occurred, the crystal orientation of the grains varied. Moreover, the direction of cold-rolling in each stage also hardly affected the texture evolution during cold-rolling and subsequent short-term annealing. During long-term annealing, sheets with the same cold-rolling direction in the as-received state and in the first stage showed the texture evolution of conventional one-way cold-rolled pure iron. Additionally, we conducted a machine learning-aided analysis of rolling and recrystallization textures. Using cold-rolling and annealing conditions as the input data and the degree of Goss orientation development as the output data, we constructed high-accuracy regression models using artificial neural networks and XGBoost. We also revealed that the annealing temperature is the dominant factor in the nucleation of Goss grains.

An Effective Federated Object Detection Framework with Dynamic Differential Privacy

The proliferation of data across multiple domains necessitates the adoption of machine learning models that respect user privacy and data security, particularly in sensitive scenarios like surveillance and medical imaging. Federated learning (FL) offers a promising solution by decentralizing the learning process, allowing multiple participants to collaboratively train a model without sharing their data. However, when applied to complex tasks such as object detection, standard FL frameworks can fall short in balancing the dual demands of high accuracy and stringent privacy. This paper introduces a sophisticated federated object detection framework that incorporates advanced differential privacy mechanisms to enhance privacy protection. Our framework is designed to work effectively across heterogeneous and potentially large-scale datasets, characteristic of real-world environments. It integrates a novel adaptive differential privacy model that strategically adjusts the noise scale during the training process based on the sensitivity of the features being learned and the progression of the model’s accuracy. We present a detailed methodology that includes a privacy budget management system, which optimally allocates and tracks privacy expenditure throughout training cycles. Additionally, our approach employs a hybrid model aggregation technique that not only ensures robust privacy guarantees but also mitigates the degradation of object detection performance typically associated with DP. The effectiveness of our framework is demonstrated through extensive experiments on multiple benchmark datasets, including COCO and PASCAL VOC. Our results show that our framework not only adheres to strict DP standards but also achieves near-state-of-the-art object detection performance, underscoring its practical applicability. For example, in some settings, our method can lower the privacy success rate by 40% while maintaining high model accuracy. This study makes significant strides in advancing the field of privacy-preserving machine learning, especially in applications where user privacy cannot be compromised. The proposed framework sets a new benchmark for implementing federated learning in complex, privacy-sensitive tasks and opens avenues for future research in secure, decentralized machine learning technologies.

Generation and Countermeasures of adversarial examples on vision: a survey

Recent studies have found that deep learning models are vulnerable to adversarial examples, demonstrating that applying a certain imperceptible perturbation on clean examples can effectively deceive the well-trained and high-accuracy deep learning models. Moreover, the adversarial examples can achieve a considerable level of certainty with the attacked label. In contrast, human could barely discern the difference between clean and adversarial examples, which raised tremendous concern about robust and trustworthy deep learning techniques. In this survey, we reviewed the existence, generation, and countermeasures of adversarial examples in Computer Vision, to provide comprehensive coverage of the field with an intuitive understanding of the mechanisms and summarized the strengths, weaknesses, and major challenges. We hope this effort will ignite further interest in the community to solve current challenges and explore this fundamental area.

Advancing Forest-Fire Management: Exploring Sensor Networks, Data Mining Techniques, and SVM Algorithm for Prediction

Forest-fire is a pressing global problem that has far-reaching effects on human life and the environment, with climate change exacerbating their frequency and intensity. There is an urgent need for advanced predictive systems to mitigate these impacts. To address this issue, this study introduces a forest-fire prediction framework integrating wireless sensor networks (WSNs), data analysis, and machine learning. Sensor nodes deployed in a forest area collected real-time meteorological data, which was transmitted using LoRaWAN technology. Data mining techniques prepared the data for analysis using the SVM algorithm, revealing relationships between meteorological parameters and wildfire risk. The SVM model demonstrated an accuracy of 86% in classifying forest-fire risk levels based on temperature, humidity, wind speed, and rainfall data. The integrated framework of WSNs and the SVM algorithm provides a high-accuracy model for forest-fire risk prediction. The model is compared to the Canadian Forest Fire Hazard Rating System to validate its accuracy, demonstrating strong agreement with historical records and reports. The model's practical implications include efficient management, early detection, and prevention strategies. However, the model's limitations suggest avenues for future research, we should consider broader geographic applications and using advanced machine-learning methods to enhance the model's predictive capabilities.

Advancing battery safety: Integrating multiphysics and machine learning for thermal runaway prediction in lithium-ion battery module

The safety concerns associated with lithium-ion batteries (LIBs) have led to the development of a novel framework combining advanced machine learning (ML) techniques with multiphysics modeling. Herein, we report an ML framework aiming to predict the occurrence of thermal runaway (TR) in the LIB module by employing a multiphysics model that incorporates thermal, electrochemical, and degradation sub-models. The focus of this research lies in understanding the degradation phenomenon associated with the breakdown of the solid electrolyte interface (SEI) on the negative electrode, which can trigger TR. The developed multiphysics model enables the investigation of electrochemical and degradation processes within batteries under various conditions, including constant charge/discharge and driving cycles. To capture the spatio-temporal temperature change, a graph neural network (GNN) for spatial change is coupled with a Long Short-Term Memory (LSTM) network for temporal evolution to form an integrated framework. The results demonstrate the high accuracy of the ML model in predicting battery temperatures in a module based on spatial and temporal temperature data obtained from temperature sensors attached to the batteries, hence, offering a means to detect TR before it occurs by identifying potential thermal hotspots.

Ethics-aware face recognition aided by synthetic face images

Current face recognition models trained on large-scale face datasets have achieved promising performance. However, using face images to train a face recognition model without consent would lead to severe privacy and ethical issues. Moreover, existing face recognition models also exhibit uneven performance on different races, thus perplexing vulnerable populations. To address the aforementioned two issues, this work investigates an ethics-aware face recognition method and examines whether we can leverage synthesized faces to achieve a high-accuracy racial balanced recognition model. In a nutshell, we introduce a race-controllable and identity-innumerable face synthesis approach to generate synthetic face images, and then employ the synthesized images to improve face recognition accuracy and mitigate recognition imbalance among different races despite the scarcity of consenting images (less than 100 individuals). More importantly, the synthetic data enable us to analyze the potential impacts of races on face recognition models quantitatively and facilitate the eradication of racial imbalance in face recognition. Extensive experiments demonstrate that employing our synthetic face data improves face recognition accuracy by a large margin while mitigating the recognition imbalance across different race groups.

An explainable neural network integrating Jiles-Atherton and nonlinear auto-regressive exogenous models for modeling universal hysteresis

The inherent nonlinear and memory-dependent input-output characteristics of piezoelectric actuators pose challenges to the precision of piezoelectric positioning systems. In order to solve this problem, this paper firstly transforms the Jiles-Atherton (JA) model into a neural network structure, designs the Jiles-Atherton neural network (JANN), and combines JANN with nonlinear autoregressive exogenous input (NARX) neural network. A hybrid JA-NARX neural network model is proposed for the first time. This model has the advantages of simple structure, high modeling accuracy, and good interpretability. The effectiveness of the proposed JA-NARX neural network model is validated through a series of experiments, specifically assessing its capacity to accurately capture rate-dependent and asymmetric hysteresis characteristics. The results show that although the proposed neural network model has fewer layers and relatively simple structure, it can realize the high-precision modeling of piezoelectric hysteresis dynamics at a lower computational cost. The experimental data shows that, under the excitation of 60 Hz input signal, the model's PV error only accounts for 0.82% of the full scale range, and the modeling performance is far superior to other models.

Estimation of winter wheat yield in Hebi city based on crop models and remote sensing data assimilation

Abstract Assimilating remote sensing data with crop models is an effective approach to improve the accuracy of crop model applications at the regional scale. Sobol global sensitivity analysis and Markov Chain Monte Carlo methods were used to calibrate parameters in the WOFOST and WheatSM models within a gridded model framework in the study area. Time series data reconstruction techniques were employed to correct MODIS LAI and ET data. Four assimilation scenarios were tested, including assimilation of only leaf area index (LAI), assimilation of only evapotranspiration (ET), simultaneous assimilation of LAI and ET and a control scenario without assimilation. These scenarios were applied to model winter wheat yields in Hebi City, China, from 2013 to 2018. Statistical validation demonstrated that assimilating ET or LAI individually significantly improved model accuracy compared to the control scenario with similar levels of improvement. The highest model accuracy was achieved when assimilating both ET and LAI simultaneously, showing the highest correlation coefficients and the lowest root mean square errors among the four scenarios. This research provides a basis for selecting assimilation variables when applying crop models at the regional scale. The coupling of crop growth models with remote sensing data empowers governments and agricultural producers to devise more effective agricultural strategies, allocate resources efficiently, and implement disaster response measures. This enhances scientific management within the agricultural sector, promotes increased food production and elevates farmers’ incomes.

Enhancing Clinical Diagnosis With Convolutional Neural Networks: Developing High-Accuracy Deep Learning Models for Differentiating Thoracic Pathologies.

Background The use of computational technology in medicine has allowed for an increase in the accuracy of clinical diagnosis, reducing errors through additional layers of oversight. Artificial intelligence technologies present the potential to further augment and expedite the accuracy, quality, and efficiency at which diagnosis can be made when used as an adjunctive tool. Such techniques, if found to be accurate and reliable in their diagnostic acuity, can be implemented to foster better clinical decision-making, improving patient quality of care while reducing healthcare costs. Methodology This study implemented convolution neural networks to develop a deep learning model capable of differentiating normal chest X-rays from those indicating pneumonia, tuberculosis, cardiomegaly, and COVID-19. There were 3,063 normal chest X-rays, 3,098 pneumonia chest X-rays, 2,920 COVID-19 chest X-rays, 2,214 chest X-rays, and 554 tuberculosis chest X-rays from Kaggle that were used for training and validation.The model was trained to recognize patterns within the chest X-rays to efficiently recognize these diseases within patients to be treated on time. Results The results indicated a success rate of 98.34% incorrect detections, exemplifying a high degree of accuracy. There are limitations to this study. Training models require hundreds to thousands of samples, and due to potential variability in image scanning equipment and techniques from which the images are sourced, the model could have learned to interpret external noise and unintended details which can adversely impact accuracy. Conclusions Further studies that implement more universal database-sourced images with similar image scanning techniques, assess diverse but related medical conditions, and the utilization of repeat trials can help assess the reliability of the model.These results highlight the potential of machine learning algorithms for disease detection with chest X-rays.

Generating synthetic building electrical load profiles using machine learning based on the CRISP-ML(Q) framework

Abstract This study presents a machine learning application for generating synthetic building electrical load profiles. The implementation followed the Cross Industry Standard Process for the development of Machine Learning Applications with Quality assurance methodology, or CRISP-ML(Q) framework, to ensure a systematic machine learning development process. The model training performance was evaluated using the mean absolute error (MAE), root mean squared error (RSME), and coefficient of determination (R2) which were observed to be 0.0739, 0.1119, and 0.5728, respectively. These metrics remained consistent during the model testing phase, suggesting robust model performance. During the initial simulation experiment, the MAE and RMSE of the generated synthetic load profile were found to be 0.014 and 0.016, respectively, underscoring high model accuracy. Additional evaluation experiments showed that the developed machine learning application can generate realistic building load profiles using high-level parameters such as building type, average daily load, and peak demand. This study can aid in the development of demand-side management strategies and building energy management systems by providing realistic building electrical load profiles especially when real-world data is limited. For future work, researchers can consider integrating additional model features, refining data processing methods, and developing an agile version of the CRISP-ML(Q) framework.

Rethinking Pretraining as a Bridge From ANNs to SNNs.

Spiking neural networks (SNNs) are known as typical kinds of brain-inspired models with their unique features of rich neuronal dynamics, diverse coding schemes, and low power consumption properties. How to obtain a high-accuracy model has always been the main challenge in the field of SNN. Currently, there are two mainstream methods, i.e., obtaining a converted SNN through converting a well-trained artificial NN (ANN) to its SNN counterpart or training an SNN directly. However, the inference time of a converted SNN is too long, while SNN training is generally very costly and inefficient. In this work, a new SNN training paradigm is proposed by combining the concepts of the two different training methods with the help of the pretrain technique and BP-based deep SNN training mechanism. We believe that the proposed paradigm is a more efficient pipeline for training SNNs. The pipeline includes pipe-S for static data transfer tasks and pipe-D for dynamic data transfer tasks. State-of-the-art (SOTA) results are obtained in a large-scale event-driven dataset ES-ImageNet. For training acceleration, we achieve the same (or higher) best accuracy as similar leaky-integrate-and-fire (LIF)-SNNs using 1/8 training time on ImageNet-1K and 1/2 training time on ES-ImageNet and also provide a time-accuracy benchmark for a new dataset ES-UCF101. These experimental results reveal the similarity of the functions of parameters between ANNs and SNNs and also demonstrate various potential applications of this SNN training pipeline.