Search

In addition to content knowledge, it is necessary to teach learning skills. Using relevant learning skills assures better learning outcomes and abilities for the whole life course. Due to a packed curriculum or teachers having insufficient skills, student learning skills are often neglected. The aim of this study is to find out whether it is possible to support the development of declarative knowledge in learning skills in teachers through procedural knowledge acquired via practical teaching activities. A quasi-experimental study using semi-structured pre- and post-intervention interviews and six-month intervention activities was conducted with 14 teachers. The procedural knowledge of the teachers was assessed based on their practical experience with the intervention activities. Their declarative knowledge was assessed based on pre- and post-intervention interview reflections. The findings revealed that procedural knowledge, being predominantly teacher centred, mostly appears through experience and reflections on the teaching practice that guides them. It also appeared that the intervention focusing on advancing procedural knowledge among teachers advanced their declarative knowledge in terms of metacognition, revealing new aspects of motivation and sense-giving for their learning. As follow-up research, we plan to continue studying conditional knowledge in teachers in order to identify situations where this is applied. [ABSTRACT FROM AUTHOR]

Coaching and mentoring has been regarded as one of the key learning techniques in the modern learning environment in the recent past. With the swift shift in the higher education sector to a more digitally driven learning environment, limited research has evaluated the impact of coaching and mentoring on student engagement in online learning. This study aimed to evaluate the impact of coaching and mentoring on enhancing student engagement in the higher education sector. The study applied a combination of quantitative and qualitative methods of data collection, where the primary data were collected from 120 respondents via a questionnaire and 20 in-depth interviews, covering undergraduates through a simple random sampling technique. The data analysis showed that the mentor's ability to facilitate learning and encourage mentees to be focused on their goals were the most significant factors that affected student engagement in online learning. In total, 43% of the respondents extremely agreed that they are willing to recommend their colleagues to receive the service of a mentor to help them reach their optimum potential in an online learning environment. The findings of the study recommend that the higher educational institutes should administer a sound mentoring process that meets the ethical backgrounds to consistently support the continuous improvement of the students in an online learning environment to enhance their engagement in learning activities. [ABSTRACT FROM AUTHOR]

Building information modeling (BIM) can theoretically facilitate collaboration among diverse design participants in construction projects, but in practice, its implementation tends to prolong the design period. Existing literature has examined some technical and managerial causes of this problem but still lacks an overall coverage of related factors. This study aims to identify the comprehensive factors affecting the teamwork efficiency in China's BIM-based collaborative design, and to investigate the critical factors and their interactions. Based on the input-process-output theory, this study initially established a hypothetical model. Potential factors were further identified through the literature review and semi-structured interviews. Questionnaire survey was conducted, and structural equation modeling was used for analysis. The results indicated that the team cooperation atmosphere is the most significant factor, followed by the collaborators' learning ability, comfort of the working environment, BIM software function, and the characteristics and arrangement of the design task (CADT). Besides, the CADT negatively affects the teamwork efficiency through the human interaction process, while other factors exert positive impacts by affecting both the personal work process and the human interaction process. The findings can help design units to determine the management focus of BIM-based collaborative design and prioritize the allocation of limited resources accordingly to maximize teamwork efficiency. [ABSTRACT FROM AUTHOR]

Deep neural networks have underpinned much of the recent progress in the field of hyperspectral image (HSI) classification owing to their powerful ability to learn discriminative features. However, training a deep neural network often requires the availability of a large number of labeled samples to mitigate over-fitting, and these labeled samples are not always available in practical applications. To adapt the deep neural network-based HSI classification approach to cases in which only a very limited number of labeled samples (i.e., few or even only one labeled sample) are provided, we propose a novel few-shot deep learning framework for HSI classification. In order to mitigate over-fitting, the framework borrows supervision from an auxiliary set of unlabeled samples with soft pseudo-labels to assist the training of the feature extractor on few labeled samples. By considering each labeled sample as a reference agent, the soft pseudo-label is assigned by computing the distances between the unlabeled sample and all agents. To demonstrate the effectiveness of the proposed method, we evaluate it on three benchmark HSI classification datasets. The results indicate that our method achieves better performance relative to existing competitors in few-shot and one-shot settings. [ABSTRACT FROM AUTHOR]

Many theories have been put forward that propose that developmental dyslexia is caused by low-level neural, cognitive, or perceptual deficits. For example, statistical learning is a cognitive mechanism that allows the learner to detect a probabilistic pattern in a stream of stimuli and to generalise the knowledge of this pattern to similar stimuli. The link between statistical learning and reading ability is indirect, with intermediate skills, such as knowledge of frequently co-occurring letters, likely being causally dependent on statistical learning skills and, in turn, causing individual variation in reading ability. We discuss theoretical issues regarding what a link between statistical learning and reading ability actually means and review the evidence for such a deficit. We then describe and simulate the "noisy chain hypothesis", where each intermediary link between a proposed cause and the end-state of reading ability reduces the correlation coefficient between the low-level deficit and the end-state outcome of reading. We draw the following conclusions: (1) Empirically, there is evidence for a correlation between statistical learning ability and reading ability, but there is no evidence to suggest that this relationship is causal, (2) theoretically, focussing on a complete causal chain between a distal cause and developmental dyslexia, rather than the two endpoints of the distal cause and reading ability only, is necessary for understanding the underlying processes, (3) statistically, the indirect nature of the link between statistical learning and reading ability means that the magnitude of the correlation is diluted by other influencing variables, yielding most studies to date underpowered, and (4) practically, it is unclear what can be gained from invoking the concept of statistical learning in teaching children to read. [ABSTRACT FROM AUTHOR]

Single-image super-resolution (SISR) techniques have been developed rapidly with the remarkable progress of convolutional neural networks (CNNs). The previous CNNs-based SISR techniques mainly focus on the network design while ignoring the interactions and interdependencies between different dimensions of the features in the middle layers, consequently hindering the powerful learning ability of CNNs. In order to address this problem effectively, a residual triplet attention network (RTAN) for efficient interactions of the feature information is proposed. Specifically, we develop an innovative multiple-nested residual group (MNRG) structure to improve the learning ability for extracting the high-frequency information and train a deeper and more stable network. Furthermore, we present a novel lightweight residual triplet attention module (RTAM) to obtain the cross-dimensional attention weights of the features. The RTAM combines two cross-dimensional interaction blocks (CDIBs) and one spatial attention block (SAB) base on the residual module. Therefore, the RTAM is not only capable of capturing the cross-dimensional interactions and interdependencies of the features, but also utilizing the spatial information of the features. The simulation results and analysis show the superiority of the proposed RTAN over the state-of-the-art SISR networks in terms of both evaluation metrics and visual results. [ABSTRACT FROM AUTHOR]

Determining ingestive behaviors of dairy cows is critical to evaluate their productivity and health status. The objectives of this research were to (1) develop the relationship between forage species/heights and sound characteristics of three different ingestive behaviors (bites, chews, and chew-bites); (2) comparatively evaluate three deep learning models and optimization strategies for classifying the three behaviors; and (3) examine the ability of deep learning modeling for classifying the three ingestive behaviors under various forage characteristics. The results show that the amplitude and duration of the bite, chew, and chew-bite sounds were mostly larger for tall forages (tall fescue and alfalfa) compared to their counterparts. The long short-term memory network using a filtered dataset with balanced duration and imbalanced audio files offered better performance than its counterparts. The best classification performance was over 0.93, and the best and poorest performance difference was 0.4–0.5 under different forage species and heights. In conclusion, the deep learning technique could classify the dairy cow ingestive behaviors but was unable to differentiate between them under some forage characteristics using acoustic signals. Thus, while the developed tool is useful to support precision dairy cow management, it requires further improvement. [ABSTRACT FROM AUTHOR]

What does the way that autistic individuals bypass, learn, and eventually master language tell us about humans' genetically encoded linguistic ability? In this theoretical review, we argue that autistic non-social acquisition of language and autistic savant abilities provide a strong argument for an innate, human-specific orientation towards (and mastery of) complex embedded structures. Autistic non-social language learning may represent a widening of the material processed during development beyond oral language. The structure detection and manipulation and generative production of non-linguistic embedded and chained material (savant abilities in calendar calculation, musical composition, musical interpretation, and three-dimensional drawing) may thus represent an application of such innate mechanisms to non-standard materials. Typical language learning through exposure to the child's mother tongue may represent but one of many possible achievements of the same capacity. The deviation from typical language development in autism may ultimately allow access to oral language, sometimes in its most elaborate forms, and also explain the possibility of the absence of its development when applied exclusively to non-linguistic structured material. Such an extension of human capacities beyond or in parallel to their usual limits call into question what we consider to be specific or expected in humans and therefore does not necessarily represent a genetic "error". Regardless of the adaptive success or failure of non-social language learning, it is the duty of science and ethical principles to strive to maintain autism as a human potentiality to further foster our vision of a plural society. [ABSTRACT FROM AUTHOR]

Simple Summary: Elephants are known for their complex vocalization system and for being able to imitate sounds. Here, we show that African elephants apply unusual and individualistic sound production mechanisms to generate idiosyncratic sounds. These sounds are produced by manipulating non-phonatory structures, e.g., applying an ingressive airflow at the trunk tip to emit extraordinarily high-frequency sounds or repeatedly contract superficial muscles at the trunk base to generate lower-frequency pulsated sounds. Intriguingly, each individual establishes its own distinctive sound-producing strategy (e.g., contracting different muscle bundles). The production of these sounds on cue is encouraged via positive reinforcement training. This suggests that social feedback and reinforcement can facilitate vocal creativity and learning behavior in elephants. Social interactions and positive feedback are also crucial for early speech learning in human infants. Increasing knowledge on sound production plasticity in elephants—long-living, highly social mammals—is crucial in the effort to better understand their communicative and vocal learning ability and its function in wild elephant populations. How do elephants achieve their enormous vocal flexibility when communicating, imitating or creating idiosyncratic sounds? The mechanisms that underpin this trait combine motoric abilities with vocal learning processes. We demonstrate the unusual production techniques used by five African savanna elephants to create idiosyncratic sounds, which they learn to produce on cue by positive reinforcement training. The elephants generate these sounds by applying nasal tissue vibration via an ingressive airflow at the trunk tip, or by contracting defined superficial muscles at the trunk base. While the production mechanisms of the individuals performing the same sound categories are similar, they do vary in fine-tuning, revealing that each individual has its own specific sound-producing strategy. This plasticity reflects the creative and cognitive abilities associated with 'vocal' learning processes. The fact that these sounds were reinforced and cue-stimulated suggests that social feedback and positive reinforcement can facilitate vocal creativity and vocal learning behavior in elephants. Revealing the mechanism and the capacity for vocal learning and sound creativity is fundamental to understanding the eloquence within the elephants' communication system. This also helps to understand the evolution of human language and of open-ended vocal systems, which build upon similar cognitive processes. [ABSTRACT FROM AUTHOR]

The classroom closure during the first semester of 2020 entailed decisive changes in higher education. Universities have become more digital in both the availability of e-resources and pervasive devices and how students communicate with lecturers and classmates. Learners adapted their study habits with a growing role of self-paced, internet-based strategies. Some flipped learning approaches have proven their efficacy under the remote-teaching physical constraints. This study aimed to appraise the outcomes from the implementation of various web-based, learning-aid tools on flipped teaching approaches in engineering modules. The open educational resources (OER) performed satisfactorily during the lockdown period in three universities from two countries with similar higher education models. Such resources encompassed classroom response systems and web-based exercise repositories, designed for diverse purposes such as autonomous learning, self-correction, flipped classroom, peer assessment, and guided study. The acquired experiences reveal that OER helped students to enhance their engagement, reach the deeper levels of the cone of learning, and widen their range of learning abilities. This procedure is easily attainable for architecture, engineering, and construction (AEC) courses and lifelong learning settings. Feedback from students, instructors' perceptions, and learning outcomes show the suitability and effectiveness of the web-based learning assistant procedure presented here. [ABSTRACT FROM AUTHOR]

Students often have difficulty understanding abstract concepts in chemistry and a low spatial ability, especially in visualizing intermolecular interactions at the submicroscopic level. Therefore, this study aims to analyze the spatial ability of students by using a 3D virtual representation as they study the curriculum topic of molecular geometry. The research was conducted with 13 male and 23 female year 10 secondary students in Jakarta, Indonesia. A qualitative methodology was employed to analyze students' spatial abilities while they undertook learning activities using 3D virtual representation. Data collection included a spatial assessment, a quiz about prior knowledge, student worksheets, interviews, observation sheets, reflective journals, and a molecular geometry test. Learning was carried out in three stages; engage, explore, and explain. The results of the research show that students' ability in spatial orientation is low whereas their ability in spatial relationships that involve using mental manipulation and rotation of 2D or 3D objects is more developed. Students employ this approach to develop their understanding of molecular geometry. [ABSTRACT FROM AUTHOR]

Designing novel robots that can cope with a specific task is a challenging problem because of the enormous design space that involves both morphological structures and control mechanisms. To this end, we present a computational method for automating the design of modular robots. Our method employs a genetic algorithm to evolve robotic structures as an outer optimization, and it applies a reinforcement learning algorithm to each candidate structure to train its behavior and evaluate its potential learning ability as an inner optimization. The size of the design space is reduced significantly by evolving only the robotic structure and by performing behavioral optimization using a separate training algorithm compared to that when both the structure and behavior are evolved simultaneously. Mutual dependence between evolution and learning is achieved by regarding the mean cumulative rewards of a candidate structure in the reinforcement learning as its fitness in the genetic algorithm. Therefore, our method searches for prospective robotic structures that can potentially lead to near-optimal behaviors if trained sufficiently. We demonstrate the usefulness of our method through several effective design results that were automatically generated in the process of experimenting with actual modular robotics kit. [ABSTRACT FROM AUTHOR]

At present, the one-stage detector based on the lightweight model can achieve real-time speed, but the detection performance is challenging. To enhance the discriminability and robustness of the model extraction features and improve the detector's detection performance for small objects, we propose two modules in this work. First, we propose a receptive field enhancement method, referred to as adaptive receptive field fusion (ARFF). It enhances the model's feature representation ability by adaptively learning the fusion weights of different receptive field branches in the receptive field module. Then, we propose an enhanced up-sampling (EU) module to reduce the information loss caused by up-sampling on the feature map. Finally, we assemble ARFF and EU modules on top of YOLO v3 to build a real-time, high-precision and lightweight object detection system referred to as the ARFF-EU network. We achieve a state-of-the-art speed and accuracy trade-off on both the Pascal VOC and MS COCO data sets, reporting 83.6% AP at 37.5 FPS and 42.5% AP at 33.7 FPS, respectively. The experimental results show that our proposed ARFF and EU modules improve the detection performance of the ARFF-EU network and achieve the development of advanced, very deep detectors while maintaining real-time speed. [ABSTRACT FROM AUTHOR]

This study proposes a knowledge-based neural fuzzy controller (KNFC) for mobile robot navigation control. An effective knowledge-based cultural multi-strategy differential evolution (KCMDE) is used for adjusting the parameters of KNFC. The KNFC is applied in PIONEER 3-DX mobile robots to achieve automatic navigation and obstacle avoidance capabilities. A novel escape approach is proposed to enable robots to autonomously avoid special environments. The angle between the obstacle and robot is used and two thresholds are set to determine whether the robot entries into the special landmarks and to modify the robot behavior for avoiding dead ends. The experimental results show that the proposed KNFC based on the KCMDE algorithm has improved the learning ability and system performance by 15.59% and 79.01%, respectively, compared with the various differential evolution (DE) methods. Finally, the automatic navigation and obstacle avoidance capabilities of robots in unknown environments were verified for achieving the objective of mobile robot control. [ABSTRACT FROM AUTHOR]

Ketamine has become a popular recreational drug due to its neuronal anesthesia effect and low price. The process of learning and memory is part of the distinctive high-level neural activities in animals. We investigated the effects of subanesthetic and anesthetic doses of ketamine on the learning and memory-related signal transduction mechanisms. We used the Morris water maze test to execute rats' learning and memory ability and detected changes of Arc mRNA and Arc, cAMP-response element-binding protein (CREB), phospho-CREB (p-CREB), extracellular signal-regulated kinase (ERK), and phospho-ERK (p-ERK) protein expression in the hippocampus 10 min and 24 h after administration. Ten min after ketamine injection, the Arc gene and the protein expression levels increased in all groups; p-ERK only increased in the chronic subanesthetic dose group. After 24 h, the Arc gene and the protein expression levels of the subanesthetic dose group increased, but those of the chronic subanesthetic dose group and anesthetic dose group decreased. However, p-ERK increased in all groups. A chronic subanesthetic dose of ketamine could increase learning and memory ability through ERK, CREB, and Arc in a short time, and the high body temperature after the subanesthetic dose of ketamine injection was the main factor leading to changes in Arc. The subanesthetic dose of ketamine regulated learning and memory through ERK, CREB, and ARC 24 h after injection. [ABSTRACT FROM AUTHOR]

Oxford Nanopore sequencing is an important sequencing technology, which reads the nucleotide sequence by detecting the electrical current signal changes when DNA molecule is forced to pass through a biological nanopore. The research on signal simulation of nanopore sequencing is highly desirable for method developments of nanopore sequencing applications. To improve the simulation accuracy, we propose a novel signal simulation method based on Bi-directional Gated Recurrent Units (BiGRU). In this method, the signal processing model based on BiGRU is built to replace the traditional low-pass filter to post-process the ground-truth signal calculated by the input nucleotide sequence and nanopore sequencing pore model. Gaussian noise is then added to the filtered signal to generate the final simulated signal. This method can accurately model the relation between ground-truth signal and real-world sequencing signal through experimental sequencing data. The simulation results reveal that the proposed method utilizing the powerful learning ability of the neural network can generate the simulated signal that is closer to the real-world sequencing signal in the time and frequency domains than the existing simulation method. [ABSTRACT FROM AUTHOR]

The scene classification of a remote sensing image has been widely used in various fields as an important task of understanding the content of a remote sensing image. Specially, a high-resolution remote sensing scene contains rich information and complex content. Considering that the scene content in a remote sensing image is very tight to the spatial relationship characteristics, how to design an effective feature extraction network directly decides the quality of classification by fully mining the spatial information in a high-resolution remote sensing image. In recent years, convolutional neural networks (CNNs) have achieved excellent performance in remote sensing image classification, especially the residual dense network (RDN) as one of the representative networks of CNN, which shows a stronger feature learning ability as it fully utilizes all the convolutional layer information. Therefore, we design an RDN based on channel-spatial attention for scene classification of a high-resolution remote sensing image. First, multi-layer convolutional features are fused with residual dense blocks. Then, a channel-spatial attention module is added to obtain more effective feature representation. Finally, softmax classifier is applied to classify the scene after adopting data augmentation strategy for meeting the training requirements of the network parameters. Five experiments are conducted on the UC Merced Land-Use Dataset (UCM) and Aerial Image Dataset (AID), and the competitive results demonstrate that our method can extract more effective features and is more conducive to classifying a scene. [ABSTRACT FROM AUTHOR]

Due to a good ability of learning for nonlinear uncertainties, a mixed modified recurring Rogers-Szego polynomials neural network (MMRRSPNN) control with mended grey wolf optimization (MGWO) by using two linear adjusted factors is proposed to the six-phase induction motor (SIM) expelling continuously variable transmission (CVT) organized system for acquiring better control performance. The control system can execute MRRSPNN control with a fitted learning rule, and repay control with an evaluated rule. In the light of the Lyapunov stability theorem, the fitted learning rule in the MRRSPNN control can be derived, and the evaluated rule of the repay control can be originated. Besides, the MGWO by using two linear adjusted factors yields two changeable learning rates for two parameters to find two ideal values and to speed-up convergence of weights. Experimental results in comparisons with some control systems are demonstrated to confirm that the proposed control system can achieve better control performance. [ABSTRACT FROM AUTHOR]

A controlled suspension usually consists of a high-level and a low-level controller. The purpose the high-level controller is to analyze external data on vehicle conditions and make decisions on the required value of the force on the shock absorber rod, while the purpose of the low-level controller is to ensure the implementation of the desired force value by controlling the actuators. Many works have focused on the design of high-level controllers of active suspensions, in which it is considered that the shock absorber can instantly and absolutely accurately implement a given control input. However, active shock absorbers are complex systems that have hysteresis. In addition, electro-pneumatic and hydraulic elements are often used in the design, which have a long response time and often low accuracy. The application of methods of control theory in such systems is often difficult due to the complexity of constructing their mathematical models. In this article, the authors propose an effective low-level controller for an active shock absorber based on a time-delay neural network. Neural networks in this case show good learning ability. The low-level controller is implemented in a simplified suspension model and the simulation results are presented for a number of typical cases. [ABSTRACT FROM AUTHOR]

The dissemination of multidrug-resistant Gram-negative bacteria (MDR-GNB) is associated with increased morbidity and mortality in several countries. Machine learning (ML) is a branch of artificial intelligence that consists of conferring on computers the ability to learn from data. In this narrative review, we discuss three existing examples of the application of ML algorithms for assessing three different types of risk: (i) the risk of developing a MDR-GNB infection, (ii) the risk of MDR-GNB etiology in patients with an already clinically evident infection, and (iii) the risk of anticipating the emergence of MDR in GNB through the misuse of antibiotics. In the next few years, we expect to witness an increasingly large number of research studies perfecting the application of ML techniques in the field of MDR-GNB infections. Very importantly, this cannot be separated from the availability of a continuously refined and updated ethical framework allowing an appropriate use of the large datasets of medical data needed to build efficient ML-based support systems that could be shared through appropriate standard infrastructures. [ABSTRACT FROM AUTHOR]

Children's creative imagination is tested through tasks involving narrative and drawing abilities for participants between the age of 8 and 12 years. The test determines the relative importance of 'narrative' against 'graphic' imagination in interpretive, problem-solving strategies, and also considers how such distinctive functions of the creative imagination could affect 'general' creativity of the child learner. Participants were chosen from designated primary schools in the state of Guanajuato, Mexico. The test on creativity complements facts from observational methodology in a population of mixed Castilian-speaking children. The name of the test is Prueba de Imaginación Creativa Niños (2008) or 'Test of Creative Imagination in Children', the Castilian acronym being PIC-N. It comprised four sub-tests: Three designed to evaluate narrative (verbal) creativity, and one for drawing (i.e., graphic) creativity. The first three 'exercises' in the suite indicates (a) fluency, (b) flexibility, and (c) originality in narrative representations, whereas the fourth indexes (d) graphic abilities of the child learner. Results suggest that creative imagination causes variations in specific aspects of creativity, like narrative and graphic improvisation, and also modifies 'general' creativity as understood from the perspective of a developmental psychology of learning abilities in growing children within the defined age group. [ABSTRACT FROM AUTHOR]

Simple Summary: Impulsivity is believed to play a role in problematic behaviors in dogs. In this study, we developed a test to assess dogs' tendency to make impulsive choices, that is their preference for smaller immediate reward instead of larger, but harder to obtain ones. Dogs were first trained that a bowl presented on a certain side always contained a large food amount, whereas the one presented on the opposite side (although at the same distance from the dog) contained less food. Then, the bowl with less food was progressively placed closer to the dog. As expected, dogs' choices to feed from the bowl with less food increased as the distance of the latter decreased. Choices did not depend on factors that could interfere, such as dogs' level of motivation for food, training experience, or learning ability. This indicates that the test is likely to be actually assessing impulsivity, not other traits. Also, female dogs were more likely to make impulsive choices than males, in accordance with what is known in humans and rodents, supporting the validity of the test. The test was completed in less than 1 h, making it a valid option to assess impulsivity in dogs in various contexts. Impulsive choices reflect an individual's tendency to prefer a smaller immediate reward over a larger delayed one. Here, we have developed a behavioural test which can be easily applied to assess impulsive choices in dogs. Dogs were trained to associate one of two equidistant locations with a larger food amount when a smaller amount was presented in the other location, then the smaller amount was placed systematically closer to the dog. Choices of the smaller amount, as a function of distance, were considered a measure of the dog's tendency to make impulsive choices. All dogs (N = 48) passed the learning phase and completed the entire assessment in under 1 h. Choice of the smaller food amount increased as this was placed closer to the dog. Choices were independent from food motivation, past training, and speed of learning the training phase; supporting the specificity of the procedure. Females showed a higher probability of making impulsive choices, in agreement with analogue sex differences found in human and rodent studies, and supporting the external validity of our assessment. Overall, the findings support the practical applicability and represent a first indication of the validity of this method, making it suitable for investigations into impulsivity in dogs. [ABSTRACT FROM AUTHOR]

Various retinal vessel segmentation methods based on convolutional neural networks were proposed recently, and Dense U-net as a new semantic segmentation network was successfully applied to scene segmentation. Retinal vessel is tiny, and the features of retinal vessel can be learned effectively by the patch-based learning strategy. In this study, we proposed a new retinal vessel segmentation framework based on Dense U-net and the patch-based learning strategy. In the process of training, training patches were obtained by random extraction strategy, Dense U-net was adopted as a training network, and random transformation was used as a data augmentation strategy. In the process of testing, test images were divided into image patches, test patches were predicted by training model, and the segmentation result can be reconstructed by overlapping-patches sequential reconstruction strategy. This proposed method was applied to public datasets DRIVE and STARE, and retinal vessel segmentation was performed. Sensitivity (Se), specificity (Sp), accuracy (Acc), and area under each curve (AUC) were adopted as evaluation metrics to verify the effectiveness of proposed method. Compared with state-of-the-art methods including the unsupervised, supervised, and convolutional neural network (CNN) methods, the result demonstrated that our approach is competitive in these evaluation metrics. This method can obtain a better segmentation result than specialists, and has clinical application value. [ABSTRACT FROM AUTHOR]