Training Process Research Articles

Enhancing Adversarial Robustness through Stable Adversarial Training

Deep neural network models are vulnerable to attacks from adversarial methods, such as gradient attacks. Evening small perturbations can cause significant differences in their predictions. Adversarial training (AT) aims to improve the model’s adversarial robustness against gradient attacks by generating adversarial samples and optimizing the adversarial training objective function of the model. Existing methods mainly focus on improving robust accuracy, balancing natural and robust accuracy and suppressing robust overfitting. They rarely consider the AT problem from the characteristics of deep neural networks themselves, such as the stability properties under certain conditions. From a mathematical perspective, deep neural networks with stable training processes may have a better ability to suppress overfitting, as their training process is smoother and avoids sudden drops in performance. We provide a proof of the existence of Ulam stability for deep neural networks. Ulam stability not only determines the existence of the solution for an operator inequality, but it also provides an error bound between the exact and approximate solutions. The feature subspace of a deep neural network with Ulam stability can be accurately characterized and constrained by a function with special properties and a controlled error boundary constant. This restricted feature subspace leads to a more stable training process. Based on these properties, we propose an adversarial training framework called Ulam stability adversarial training (US-AT). This framework can incorporate different Ulam stability conditions and benchmark AT models, optimize the construction of the optimal feature subspace, and consistently improve the model’s robustness and training stability. US-AT is simple and easy to use, and it can be easily integrated with existing multi-class AT models, such as GradAlign and TRADES. Experimental results show that US-AT methods can consistently improve the robust accuracy and training stability of benchmark models.

PREDICTION SYSTEM FOR THE AMOUNT OF SUGAR PRODUCTION USING ADAPTIVE NEURO-FUZZY INFERENCE SYSTEM

Sugar is one of the staple foods most Indonesians use, so sugar production needs to be done optimally to meet people's needs. This research will design a prediction system for the amount of sugar production in PTPN XI PG Prajekan using the Adaptive Neuro-Fuzzy Inference System (ANFIS) method. ANFIS is a combined method of two systems, namely a fuzzy logic system and an artificial neural network system. This research consists of data collection, ANFIS system design, ANFIS training, ANFIS testing, accuracy calculation, and result analysis. The prediction system for the amount of sugar production is designed to predict the variable which is the amount of sugar production in the year using the input variables (sugarcane harvested area in year ), (amount of sugarcane in year ), (average of yield in year ), and (number of milling days in year ). The experiments in this research used variations of the type of membership function and the number of membership functions. The best model obtained in this research is a model with a difference between two sigmoidal membership functions and a product of two sigmoidal membership functions with a total of 2 membership functions for each input variable. Both models have the same Mean Absolute Percentage Error (MAPE) value, which is 1.79% in the training process and 4.82% in the testing process.

Deep reinforcement learning augmented decision model for intelligent driving cars

In view of the fact that the state machine decision model cannot effectively handle the rich contextual information and the influence of uncertain factors in ice and snow environments, a deep reinforcement learning agent based on the deep Q network algorithm (DQN) was constructed. The motion planner was used to augment the agent, and the rule-based decision planning module and the deep reinforcement learning model were integrated to establish the DQN-planner model, thereby improving the convergence speed and driving ability of the reinforcement learning agent. Finally, based on the CARLA simulation platform, a comparative experiment was conducted on the driving ability of the DQN model and the DQN-planner model on low-adhesion ice and snow roads, and the training process and verification results were analyzed respectively.

Retrieval of Solar Shortwave Irradiance from All-Sky Camera Images

The present work proposes a new model based on a convolutional neural network (CNN) to retrieve solar shortwave (SW) irradiance via the estimation of the cloud modification factor (CMF) from daytime sky images captured by all-sky cameras; this model is named CNN-CMF. To this end, a total of 237,669 sky images paired with SW irradiance measurements obtained by using pyranometers were selected at the following three sites: Valladolid and Izaña, Spain, and Lindenberg, Germany. This dataset was randomly split into training and testing sets, with the latter excluded from the training model in order to validate it using the same locations. Subsequently, the test dataset was compared with the corresponding SW irradiance measurements obtained by the pyranometers in scatter density plots. The linear fit shows a high determination coefficient (R2) of 0.99. Statistical analyses based on the mean bias error (MBE) values and the standard deviation (SD) of the SW irradiance differences yield results close to −2% and 9%, respectively. The MBE indicates a slight underestimation of the CNN-CMF model compared to the measurement values. After its validation, model performance was evaluated at the Antarctic station of Marambio (Argentina), a location not used in the training process. A similar comparison between the model-predicted SW irradiance and pyranometer measurements yielded R2=0.95, with an MBE of around 2% and an SD of approximately 26%. Although the precision provided by the SD at the Marambio station is lower, the MBE shows that the model’s accuracy is similar to previous results but with a slight overestimation of the SW irradiance. Finally, the determination coefficient improved to 0.99, and the MBE and SD are about 3% and 11%, respectively, when the CNN-CMF model is used to estimate daily SW irradiation values.

An Automatic Data-Driven Long-term Rainfall Prediction using Humboldt Squid Optimized Convolutional Residual Attentive Gated Circulation Model in India

<p style="text-align:justify;"><span style="font-family:"Times New Roman","serif";font-size:12pt;line-height:200%;">Precise long-term rainfall forecasting is becoming progressively significant, predominantly in periods of fluctuating weather circumstances. These prophecies are indispensable for several areas, including agronomy, water resource supervision, flood attentiveness, and effluence monitoring. This framework investigates the composite association between climatological data, with an emphasis on the precise estimating of rainfall by finding the effect of temperature disparities on rainfall outlines in diverse states of India. The developed framework undergoes two phases: pre-processing and rainfall forecasting. At first, the Improved min-max normalization (I-MMN) technique is highlighted in the preprocessing stage to eliminate unsolicited missing values from the database. Then, the convolutional residual attentive gated circulation unit (CResAtt-GCU) technique is presented to envisage the amount of rainfall that occurs in various Indian states. Finally, the Humboldt squid optimizer (HBSO) technique is presented to tune the parameters of the developed model to minimize network complications during the training process. The developed framework is processed under the Python platform and the real-time rainfall prediction database collected between 2000-2023 is utilized for the experimentation process. Various computational measures like R2, root mean square error (RMSE), Kling-Gupta Efficiency (KGE), Nash-Sutcliffe Efficiency (NSE), mean absolute error (MAE), and computation time are evaluated and distinguished from different studies. The overall R2 of 0.993, RMSE of 0.84, KGE of 0.96, NSE of 0.992, MAE of 0.53, and overall CT of 86.66s are obtained by the developed technique against various conventional studies on forecasting the rainfall in different regions of India.</span></p>

Distributed photovoltaic power forecasting based on personalized federated adversarial learning

Existing distributed photovoltaic (PV) power forecasting methods fail to address the impact of sample scarcity and heterogeneity in PV power data. Moreover, training a single prediction model proves challenging to meet the personalized forecasting needs of different PV stations in distributed environments. This paper proposes a personalized federated generative adversarial network (PFedGAN)-based DPV power forecasting method. Leveraging the federated learning (FL) framework, it achieves collaborative training of prediction models among DPV stations while preserving data privacy. y introducing generative adversarial networks (GAN) and personalized strategy optimization into the FL training process, it alleviates data scarcity issues and reduces the impact of data heterogeneity. A TimesNet-DeepAR (TNE-DeepAR) power prediction model is designed, where the TimesNet module extracts correlations between PV power data from different time periods, and the DeepAR module facilitates PV power prediction, mitigating the effects of meteorological factors' multi-periodic variations on PV power. Experimental results show that the proposed hybrid prediction model reduces the average mean absolute percentage error (MAPE) by 30–40 % compared to single models. The proposed approach reduces the MAPE by 9.79 % compared to traditional methods and by 49.62 % for PV stations with scarce data.

Simple and efficient step detection algorithm for foot-mounted IMU

Abstract This paper presents a concise, efficient, and adaptive step detection algorithm based on foot-mounted inertial measurement unit sensors. The proposed method maps the temporal values of pedestrian motion and gait diversity into two variables: the distance between peaks and valleys, and the slope. Compared to traditional sliding window methods, this approach amplifies the differences between normal and abnormal steps, allowing it to adapt to various indoor activities such as fast walking, slow walking, running, jogging, standing still, and turning. By incorporating adaptive factors, it addresses the challenge of detecting steps while going up and down stairs. The proposed algorithm overcomes the limitations of traditional adaptive threshold methods that require different temporal and peak thresholds for various gait conditions. By utilizing the significant differences in distance and slope, it effectively resolves the issue of detecting steps during stationary periods. Unlike neural network-based gait classifiers, this algorithm does not need to account for multiple gait conditions, thereby simplifying the training process. Experimental results demonstrate that the algorithm achieves an average accuracy of over 99% under mixed indoor walking conditions and over 98% accuracy in long-term outdoor walking conditions.

Saliva as a diagnostic tool in soccer: a scoping review.

A high-performance sport like soccer requires training strategies that aim to reach peak performance at the right time for the desired competitions. Thus, the investigation of biochemical markers in saliva is a tool that is beginning to be used in athletes within the physical training process. There is still no evidence on universal saliva collection and analysis protocols in soccer. This review aims to map the use of saliva as a tool for analyzing athletic performance in soccer, from the biomarkers used to the validated protocols for these analyses. A broad systematic literature search was carried out in the electronic databases Web of Science, Livivo, Scopus, PubMed, LILACS and gray literature (Google Scholar and ProQuest). Two reviewers selected the studies and extracted data on the type of salivary collection used, the salivary biomarker evaluated and monitored. Ninety-three articles were included. The most frequently analyzed salivary biomarkers were cortisol (n=53), testosterone (n=35), secretory immunoglobulin A (SIgA) (n=33), salivary alpha amylase (n=7), genetic polymorphisms (n=4) and miRNAs (n=2). The results of the studies indicated beneficial effects in monitoring salivary biomarkers in the assessment of sports performance, although most studies did not include a control group capable of comparison. Salivary collection and analysis protocols were varied and commonly not reported. This scoping review provides a comprehensive overview of the current landscape of salivary biomarker research in soccer. The findings underscore the importance of these biomarkers in assessing athletes' physiological responses and overall well-being. Future research should focus on refining methodologies, exploring additional biomarkers, and investigating the practical implications of salivary biomarker monitoring in soccer and other sports.

Styled and characteristic Peking opera facial makeup synthesis with co-training and transfer conditional styleGAN2

Against the backdrop of the deep integration of culture and technology, research and practice in digitization of intangible cultural heritage has continued to deepen. However, due to the lack of data and training, it is still very difficult to apply artificial intelligence to the field of cultural heritage protection. This article integrates image generation technology into the digital protection of Peking opera facial makeup, using a self-built Peking opera facial makeup dataset. Based on the StyleGAN2 network, we propose a style generative cooperative training network Co-StyleGAN2, which integrates the adaptive data augmentation (ADA) to alleviate the problem of discriminator overfitting and introduces the idea of cooperative training to stabilize the training process. We design a Peking opera facial makeup image transform conditional generation network TC-StyleGAN2 which is transferred from unconditional generation network. The weights of the unconditional pre-training model are fixed, and an adaptive filtering modulation module is added to modulate the category parameters to complete the conversion from unconditional to conditional StyleGAN2 to deal with the training difficulty of conditional GANs on limited data, which suffer from severe mode collapse. The experimental results show that the proposed training strategy is better than the comparison algorithms, and the image generation quality and diversity have been improved.

Early Cervical Cancer Diagnosis with SWIN-Transformer and Convolutional Neural Networks.

Introduction: Early diagnosis of cervical cancer at the precancerous stage is critical for effective treatment and improved patient outcomes. Objective: This study aims to explore the use of SWIN Transformer and Convolutional Neural Network (CNN) hybrid models combined with transfer learning to classify precancerous colposcopy images. Methods: Out of 913 images from 200 cases obtained from the Colposcopy Image Bank of the International Agency for Research on Cancer, 898 met quality standards and were classified as normal, precancerous, or cancerous based on colposcopy and histopathological findings. The cases corresponding to the 360 precancerous images, along with an equal number of normal cases, were divided into a 70/30 train-test split. The SWIN Transformer and CNN hybrid model combines the advantages of local feature extraction by CNNs with the global context modeling by SWIN Transformers, resulting in superior classification performance and a more automated process. The hybrid model approach involves enhancing image quality through preprocessing, extracting local features with CNNs, capturing the global context with the SWIN Transformer, integrating these features for classification, and refining the training process by tuning hyperparameters. Results: The trained model achieved the following classification performances on fivefold cross-validation data: a 94% Area Under the Curve (AUC), an 88% F1 score, and 87% accuracy. On two completely independent test sets, which were never seen by the model during training, the model achieved an 80% AUC, a 75% F1 score, and 75% accuracy on the first test set (precancerous vs. normal) and an 82% AUC, a 78% F1 score, and 75% accuracy on the second test set (cancer vs. normal). Conclusions: These high-performance metrics demonstrate the models' effectiveness in distinguishing precancerous from normal colposcopy images, even with modest datasets, limited data augmentation, and the smaller effect size of precancerous images compared to malignant lesions. The findings suggest that these techniques can significantly aid in the early detection of cervical cancer at the precancerous stage.

Knowledge graph-enhanced multi-agent reinforcement learning for adaptive scheduling in smart manufacturing

AbstractSelf-organizing manufacturing network has emerged as a viable solution for adaptive manufacturing control within the mass personalization paradigm. This approach involves three critical elements: system modeling and control architecture, interoperable communication, and adaptive manufacturing control. However, current research often separates interoperable communication from adaptive manufacturing control as isolated areas of study. To address this gap, this paper introduces Knowledge Graph-enhanced Multi-Agent Reinforcement Learning (MARL) method that integrates interoperable communication via Knowledge Graphs with adaptive manufacturing control through Reinforcement Learning. We hypothesize that implicit domain knowledge obtained from historical production job allocation records can guide each agent to learn more effective scheduling policies with accelerated learning rates. This is based on the premise that machine assignment preferences effectively could reduce the Reinforcement Learning search space. Specifically, we redesign machine agents with new observation, action, reward, and cooperation mechanisms considering the preference of machines, building upon our previous MARL base model. The scheduling policies are trained under extensive simulation experiments that consider manufacturing requirements. During the training process, our approach demonstrates improved training speed compared with individual Reinforcement Learning methods under the same training hyperparameters. The obtained scheduling policies generated by our Knowledge Graph-enhanced MARL also outperform both individual Reinforcement Learning methods and heuristic rules under dynamic manufacturing settings.

A Study on Enhancing the Visual Fidelity of Aviation Simulators Using WGAN-GP for Remote Sensing Image Color Correction

When implementing outside-the-window (OTW) visuals in aviation tactical simulators, maintaining terrain image color consistency is critical for enhancing pilot immersion and focus. However, due to various environmental factors, inconsistent image colors in terrain can cause visual confusion and diminish realism. To address these issues, a color correction technique based on a Wasserstein Generative Adversarial Network with Gradient Penalty (WGAN-GP) is proposed. The proposed WGAN-GP model utilizes multi-scale feature extraction and Wasserstein distance to effectively measure and adjust the color distribution difference between the input image and the reference image. This approach can preserve the texture and structural characteristics of the image while maintaining color consistency. In particular, by converting Bands 2, 3, and 4 of the BigEarthNet-S2 dataset into RGB images as the reference image and preprocessing the reference image to serve as the input image, it is demonstrated that the proposed WGAN-GP model can handle large-scale remote sensing images containing various lighting conditions and color differences. The experimental results showed that the proposed WGAN-GP model outperformed traditional methods, such as histogram matching and color transfer, and was effective in reflecting the style of the reference image to the target image while maintaining the structural elements of the target image during the training process. Quantitative analysis demonstrated that the mid-stage model achieved a PSNR of 28.93 dB and an SSIM of 0.7116, which significantly outperforms traditional methods. Furthermore, the LPIPS score was reduced to 0.3978, indicating improved perceptual similarity. This approach can contribute to improving the visual elements of the simulator to enhance pilot immersion and has the potential to significantly reduce time and costs compared to the manual methods currently used by the Republic of Korea Air Force.

Challenges of tertiary higher education in the framework of highly qualified teacher training in Ghana

The study of the features of the educational system of the Republic of Ghana in the context of the third stage of higher education is necessary for the further development of Russian-Ghanaian relations and the growth of partnership projects between higher educational institutions of Russia and Ghana. A proper understanding of the educational system of the Republic of Ghana, as well as the existing problems in the training of postgraduate students and possible ways to solve them, can help to improve the training process of highly qualified teaching staff. The purpose of the work is to reveal the problems of the third stage of higher education in the context of teacher training, and identify possible ways to solve them. The study analyses: 1) the characteristics of the higher education sector in the Republic of Ghana, in particular its tertiary level; 2) existing state initiatives in the country to improve the training of highly qualified teaching staff on the third stage of higher education; 3) statistics characterizing the current situation of postgraduate studies within the framework of higher education in the country; 4) the challenges emerging in the training of highly qualified teachers in sub-Saharan Africa, as well as their relevance to tertiary education in the Republic of Ghana. The study has identified the following difficulties that affect the process of writing and successful defense of dissertations: insufficient awareness of graduate students about further prospects for their employment with a PhD; insufficient financial security of postgraduate students; insufficient level of scientific supervision and excessive academic workload on supervisors of dissertation research. The study proposes solutions to the existing problems, such as increasing financial incentives for high-achieving PhD students, increasing control over the process of scientific supervision, placing greater emphasis on conducting interdisciplinary research and establishing partnerships with other countries of the African continent, and developing existing interaction with states outside Africa, including the Russian Federation.

Massive Education in Prison Health in Brazil: A Look Beyond the Walls

Equal access to health initiatives and services under the principles of universal and comprehensive care remains a challenge in Brazil. The realization of public health policies is further intricate when one examines the health situation of people deprived of liberty. This study showcases the “Prison System: Beyond the Walls” educational pathway, available on the Virtual Learning Environment of the Brazilian National Health System (AVASUS). The action research methodological strategy guided the pathway development, emphasizing dialogic learning. The goal was to address the need for massive training on the topic of prison health, with the model focusing on engagement through spontaneous, non-mandatory participation in the pathway courses. The pathway comprised four modules, whose educational offerings were based on the self-learning model. Students were free to choose which courses to take and in what order, as there was no prerequisite for participating in modules. Hence, students could either take all the courses or only those with which they identify their learning needs, regardless of work demands or personal interests. Structuring the pathway through action research facilitated a massive, cohesive, and continuous training process. This approach expanded knowledge and established meaningful relationships among the related topics and the key players involved: health professionals, prison officers, and people deprived of liberty. Notably, the pathway courses have surpassed the 50,000-enrollment mark, spanning all five regions of Brazil. In this context, this article presents and discusses the development of the “Prison System: Beyond the Walls” pathway, emphasizing the massive improvement of health within Brazil’s prison system and highlighting the results achieved.

Reflection of Crew Resource Management (CRM) Trainings to Real-Life Field Practice in Air Passenger Transportation: A Qualitative Research

In the air transportation sector, flight safety stands as one of the foremost values. To ensure safety, airlines rely on Crew Resource Management (CRM) training. These training programs aim to foster coordination, communication, and awareness among all personnel involved in flight operations, encouraging rational responses to potential issues as a unified team. Naturally, as is the case with any training process, the practical applicability of CRM training alongside its theoretical structure is of paramount importance. Hence, this study, conducted within the framework of qualitative research, delves into the perspectives of cabin crew members regarding CRM training by evaluating its content, benefits, and real-life field applications. Employing interview-based methodology, qualitative data were gathered from 19 cabin crew participants through purposive sampling. The collected data underwent descriptive analysis, leading to certain conclusions. The findings illustrate that CRM training has a positive impact on emotional and knowledge management, communication skills, team collaboration, and a proactive approach in matters of safety. It underscores that CRM training significantly permeates into the realm of real-world applications—namely, flight operations. However, participants highlighted certain issues and emphasized the need for more practical, interactive, and participant-centered training methods for the enhancement of CRM training, aiming to address those concerns.

Components of the hr management process of a medical organization

Successful development of any sphere of human activity is unthinkable in modern society without management. The implementation of management goals and objectives is carried out through personnel policy, which is a strategic line of behavior in working with personnel. In the healthcare system, an important role is assigned to the mid-level medical worker, therefore, personnel development management is recognized throughout the world as a priorit y task of management in the field of nursing. Target. To study and analyze the process of professional selection, training and management of nursing personnel. Results. The aspects of selection of nursing personnel were studied, the level of quality of knowledge obtained by nursing personnel and the degree of their application in practical activities were determined, the features of the organizational activities of nursing service managers were identified. Conclusion. Trained medical personnel should take a place in accordance with the needs of national health care and taking into account the personal interests and needs of the medical workers themselves. This can be achieved with the appropriate organization of the personnel management process, when specialists will find adequate use for their qualifications in the system.

PLIC-FSR-SYS: System reliability analysis based on parallel learning of influential components with filtered sample region

In practical engineering, system reliability analysis is highly concerned since many structures or products have multiple failure modes. Accordingly, this paper develops an innovative method for system reliability analysis by parallel learning of influential component limit-state functions with filtered sample region (PLIC-FSR-SYS) based on Kriging modeling. Different from the traditional adaptive learning methods that train only one component in each iteration when constructing the surrogate of the composite limit-state function, a new strategy is explored to adaptively identify several important components in one iteration so as to train them simultaneously. In the meanwhile, a filtering formula is explored to determine the fatal region so that the unimportant samples can be removed to further accelerate the training process. Based on the join forces of parallel learning of influential components and avoiding the training at unimportant samples, PLIC-FSR-SYS can achieve a fairly efficient system reliability analysis with multiple failure modes. Finally, four different case studies, including an engineering application to the ultra-voltage on-load tap-changer, are conducted to prove the effectiveness of the proposed method. The results indicate that compared to traditional adaptive learning methods, the proposed method makes a significant efficiency improvement for system reliability analysis with multiple failure modes.

Comparison of deep-learning multimodality data fusion strategies in mandibular osteoradionecrosis NTCP modelling using clinical variables and radiation dose distribution volumes

Objective.Normal tissue complication probability (NTCP) modelling is rapidly embracing deep learning (DL) methods, acknowledging the importance of spatial dose information. Finding effective ways to combine information from radiation dose distribution maps (dosiomics) and clinical data involves technical challenges and requires domain knowledge. We propose different multi-modality data fusion strategies to facilitate future DL-based NTCP studies.Approach.Early, joint and late DL multi-modality fusion strategies were compared using clinical and mandibular radiation dose distribution volumes. These were contrasted with single-modality models: a random forest trained on non-image data (clinical, demographic and dose-volume metrics) and a 3D DenseNet-40 trained on image data (mandibular dose distribution maps). The study involved a matched cohort of 92 osteoradionecrosis cases and 92 controls from a single institution.Main results.The late fusion model exhibited superior discrimination and calibration performance, while the join fusion achieved a more balanced distribution of the predicted probabilities. Discrimination performance did not significantly differ between strategies. Late fusion, though less technically complex, lacks crucial inter-modality interactions for NTCP modelling. In contrast, joint fusion, despite its complexity, resulted in a single network training process which included intra- and inter-modality interactions in its model parameter optimisation.Significance.This study is a pioneering effort in comparing different strategies for including image data into DL-based NTCP models in combination with lower dimensional data such as clinical variables. The discrimination performance of such multi-modality NTCP models and the choice of fusion strategy will depend on the distribution and quality of both types of data. Multiple data fusion strategies should be compared and reported in multi-modality NTCP modelling using DL.

Optimised prediction of tunnel fire heat release rate using the ResNet18_2CLSTM model with bagging for multimodal data

Accurate predictions of HRR will improve preparedness and response strategies, enhance safety, and minimise damage in tunnel fires. In this study, a deep learning prediction model for HRR under multimodal data fusion is proposed. A multimodal dataset is first established to obtain flame images and flue gas time series data through model-scale tunnel fire experiments. During the model training process, ResNet18 was used to extract features from the flame image, and 2CLSTM was employed to understand the time series of the flame image features and flue gas features to establish the correlation with the HRR. It was evaluated that the error analyses of the measured and predicted values of the validation set yielded R2 greater than 0.85, with errors and standard deviations less than 4 kW. And the model predicted better in the flame growth and decay phases. However, there is some deviation in the predictions near the peak HRR. To address this issue, the Bagging algorithm was introduced to optimise the model. The results show that the ResNet18_2CLSTM model with Bagging reduces the RMSE by 20.47 % and increases the R2 by 4.64 % compared to the original model, and the accuracy is greatly improved.

On the Content and Training of Chinese Character Learning Strategies

Chinese character learning is a major focus and difficulty in learning Chinese as a second language. In order to learn Chinese characters more efficiently, learners can adopt appropriate Chinese character learning strategies. The article firstly classifies the Chinese character learning strategies in second language acquisition and analyses the main influencing factors of Chinese character learning. Secondly, the target learners of Chinese characters are classified into two categories: Chinese character circle and non-Chinese character circle, and the differences of their Chinese character learning strategies are analysed. Finally, the training process of Chinese character learning strategies is analysed. The aim is to further understand the use and training of Chinese character learning strategies, and to provide help and guidance for future Chinese language teachers to train learners in Chinese character learning strategies in a targeted manner.