Use Of Deep Learning Strategies Research Articles

Need Satisfaction, Motivation, and Learning Strategies in Undergraduate Kinesiology Students

ABSTRACT The purpose of this study was to examine the influence of basic psychological need satisfaction and motivation on students’ use of deep and surface learning strategies across undergraduate kinesiology courses. This study used a survey consisting of demographic questions and three self-report questionnaires assessing basic need satisfaction, motivation, and deep and surface learning strategies. The survey was administered via Qualtrics at the end of two semesters (fall 2020 and spring 2021). The participants consisted of 1,125 undergraduate kinesiology students (81.9% female, 55.3% white, 94.8% non-Hispanic) at a midsize university in the southeastern part of the United States. Path modeling showed that intrinsic motivation was influenced by autonomy, competence, and relatedness, while identified regulation was influenced by autonomy and competence. Relatedness also influenced both introjected regulation and external regulation. Participants’ intrinsic motivation and identified regulation negatively predicted surface learning strategies while external regulation positively predicted surface learning strategies. Intrinsic motivation and identified motivation predicted deep learning strategies. Interestingly, all three basic psychological needs predicted deep learning strategies. These findings underscore the importance of satisfying basic needs to influence motivation and, in certain cases, learning strategies. While it is evident that the use of deep learning strategies would be highly beneficial for this group of learners, kinesiology students may not rely on these strategies due to a lack of intrinsic motivation and basic need satisfaction. Although future research is necessary to test the causal links between these constructs, supporting students’ basic psychological needs and fostering intrinsic motivation could contribute to the use of deeper learning strategies.

Comparative Study of Artificial Intelligence Models for Breast Cancer Detection

The most prevalent type of cancer among women is breast cancer. According to the statistics given by the World Health Organization (WHO), breast cancer is the reason behind the death of about 2.3 billion women globally in 2020, accounting for 685.9 million deaths. Since they are thought to be useful approaches, machine learning and deep learning techniques have drawn attention from researchers in breast cancer detection. Also, it can significantly assist in the process of prior detection and prediction of breast cancer by extracting handcrafted features. However, in recent years, improvements in artificial intelligence (AI) have enabled the successful use of deep learning strategies like CNN and the transfer learning method for detection of breast cancer. A significantly large dataset is used for deep learning methods. It does not require human intervention for feature extraction, which, as a result, enhances the patient's chances of survival. This review paper is based on breast cancer detection using deep learning and machine learning-based cancer detection techniques to aid in the understanding of trends and challenges in cancer detection.

Identificando el rol de la inteligencia emocional en las emociones de logro y en sus efectos sobre las estrategias de aprendizaje profundo en estudiantes universitarios

Emotional skills have been proposed as distal determinants of achievement emotions. However, the specific role they play in such emotions and their effects on learning has not been defined. The purpose of this comparative study is to provide an ex post facto assessment of emotional intelligence (EI) as a predictor of the enjoyment and anger expressed while studying and as a mediator and/or moderator of its effects on the use of deep learning strategies. A sample comprising 603 university students (85.7% women; Medad=25.16, SDedad=8.42) completed the Trait Meta-Mood State-24 to assess perceived EI, the Achievement Emotions Questionnaire-AEQ for enjoyment and anger, and the Deep Learning Strategies Questionnaire. The stepwise regression, multiple mediation, and moderation analyses conducted showed that EI explains enjoyment and anger, although its different skills predict them differently. Enjoyment increases the use of deep learning strategies, an effect that is partially mediated by emotional repair. In anger there is less use of these strategies, and this effect is partially mediated by high emotional attention with little emotional clarity and repair. Similarly, a medium and high level of EI moderates the negative effect of anger on the use of deep learning strategies. These results show the benefits of being emotionally intelligent since it mitigates the negative consequences of negative emotions, stressing the need to teach emotional skills or promote EI in educational and cognitive contexts.

Deadlines make you productive, but what do they do to your motivation? Trajectories in quantity and quality of motivation and study activities among university students as exams approach.

Recent research has emphasized that achievement motivation is context-sensitive and varies within individual students. Ubiquitous temporal landmarks such as exams or deadlines are evident contextual factors that could systematically explain variation in motivation. Indeed, research has consistently found that university students increase their study efforts as exams come closer in time, indicating increasing study motivation. However, changes in study motivation for a specific exam as it comes closer have rarely been investigated. Instead, research on developmental changes in expectancy and value beliefs has consistently founds that achievement motivation declines over a semester. Surprisingly, declining motivation thus apparently coincides with increasing study efforts for end-of-semester exams. The present research investigates this apparent contradiction by assessing how exam-specific motivation and study behavior change under equal methodological conditions as an exam draws closer. Using parallel growth curve models, we examine changes in expectancy-value beliefs, performance approach and avoidance motivation and study behavior as well as motivational want- and should-conflicts among 96 students over eight weekly measurement points. Results show that students study more for their exam as it comes closer and increase their use of surface learning strategies more rapidly than their use of deep learning strategies. However, even exam-specific expectancy and attainment value beliefs decline while performance-avoidance motivation increases over time, indicating that students increasingly study out of fear to fail as exams come closer. Consistent with these findings, students' experience of should conflicts decreases while their want conflicts increase over time. We discuss several possible mechanisms underlying our findings in addition to potential theoretical consequences and suggest future research opportunities to better understand students' changes in situative motivation and study behavior in the context of temporal landmarks.

Identifying the role of emotional intelligence in achievement emotions and their effects on deep learning strategies in university students

Emotional skills have been proposed as distal determinants of achievement emotions. However, the specific role they play in such emotions and their effects on learning has not been defined. The purpose of this comparative study is to provide an ex post facto assessment of emotional intelligence (EI) as a predictor of the enjoyment and anger expressed while studying and as a mediator and/or moderator of its effects on the use of deep learning strategies. A sample comprising 603 university students (85.7% women; Mage=25.16, SDage = 8.42) completed the Trait Meta-Mood State 24 to assess perceived EI, the Achievement Emotions Questionnaire-AEQ for enjoyment and anger, and the Deep Learning Strategies Questionnaire. The stepwise regression, multiple mediation, and moderation analyses conducted showed that EI explains enjoyment and anger, although its different skills predict them differently. Enjoyment increases the use of deep learning strategies, an effect that is partially mediated by emotional repair. In anger there is less use of these strategies, and this effect is partially mediated by high emotional attention with little emotional clarity and repair. Similarly, a medium and high level of EI moderates the negative effect of anger on the use of deep learning strategies. These results show the benefits of being emotionally intelligent since it mitigates the negative consequences of negative emotions, stressing the need to teach emotional skills or promote EI in educational and cognitive contexts.

A deep learning-based bilingual Hindi and Punjabi named entity recognition system using enhanced word embeddings

The increasing availability of information on the web makes the task of named entity recognition (NER) more challenging. Named entity recognition is an important pre-processor tool that is concerned with the extraction of entities of our interest such as person, location, organization, gene, protein, number, measurement, etc. The success of earlier named entity recognition systems is highly dependent on rule-based techniques or traditional machine learning algorithms exploiting several linguistic and non-linguistic features. In this article, we propose a novel named entity recognition (NER) system that involves the use of deep learning strategies as well as an enhanced version of word embeddings. We develop a Bidirectional Gated Recurrent Unit (Bi-GRU) and Convolutional Neural Networks (CNN) based bilingual named entity recognition system which is built upon enhanced word embeddings (EWE). Enhanced word embeddings (EWE) are generated by concatenation of FastText word embeddings along with minimal feature embeddings, namely part of speech embeddings, word prefix embeddings, word suffix embeddings, and word length embeddings which improve the computational power of deep learning methods. We perform several experiments using corpora in two different languages. One is IJCNLP-08 NERSSEAL shared task corpora containing annotated dataset in Hindi language and the other is manually annotated dataset in Punjabi language. We also make several experiments on bilingual Hindi and Punjabi dataset. The results of the experiments performed in this work reveal that the Bidirectional GRU and CNN based model along with enhanced word embeddings (EWE) has excelled with Precision, Recall, and F-score value of 92.60%, 90.70%, 91.64% respectively for Hindi, 93.87%, 93.33%, 93.60% respectively for Punjabi and 93.78%, 92.66%, 93.22% respectively for bilingual Hindi and Punjabi named entity recognition. Enhanced word embeddings accelerate the performance of a Bi-GRU and CNN based named entity recognition system without using a large set of features and any sort of gazetteers.

Estrategias de aprendizaje profundas: Validación de un modelo situacional y su cuestionario

Measuring self-regulated learning is crucial to improve our educational interventions. Self-report has been the major data collection method and a number of questionnaires exist. Importantly, the vast majority of the questionnaires are constructed from general theoretical models. Our aim was to develop a model and its questionnaire –i.e. Deep Learning Strategies questionnaire- to investigate how students regulate their learning strategies in more realistic learning situations. Four scales were created: (1) Basic learning self-regulation strategies; (2) Visual elaboration and summarizing strategies; (3) Deep information processing strategies; and (4) Social learning self-regulation strategies. A total of 601 higher education students formed the sample. We analyzed, first, the internal validity of the questionnaire. Three structural models were tested: (M1) mono-factor; (M2) scales correlate among them freely, and (M3) the scales are indicators of a general construct. The latter model showed a slight better fit. Additionally, a path analysis was carried out to study the degree in which the use of the Deep learning strategies depends on personal factors and is associated to performance. It was found that the use depends directly and positively on Learning goal orientation, on the self-messages defining the Self-regulation style of emotion and motivation focused on learning, and on Effort. Besides, these two last variables convey the effect of Self-efficacy that, at the same time, affects Effort. Academic performance depends positively on Effort but negatively to the use of Deep learning strategies. It is hypothesized this negative relationship is due to the method of measurement of academic performance.

Deep Learning techniques for effective diagnosis of Alzheimer's disease using MRI images

The determination of Alzheimer’s disease (AD) from neuroimaging data such as MRI has been immensely researched over the last few years. Recent advances in deep learning from a computer perspective have advanced in that research. However, the general limitations of such algorithms depend on the large number of training images, as well as the need to carefully perform the construction of deep networks. In past few days deep learning strategies have found great achievement in the analysis of medical imaging. But very little has been done in the use of deep learning strategies to turn up and differentiate Alzheimer’s disease. We are building a deep convolutional network and demonstrating performance on the ADNI-Alzheimer’s Disease Neuroimaging Initiative Dataset. We present a deep convolutional neural network to recognize Alzheimer and differentiate according the current phase of the disease.

Deep learning self-regulation strategies: Validation of a situational model and its questionnaire

Measuring self-regulated learning is crucial to improve our educational interventions. Self-report has been the major data collection method and a number of questionnaires exist. Importantly, the vast majority of the questionnaires are constructed from general theoretical models. Our aim was to develop a model and its questionnaire –i.e. Deep Learning Strategies questionnaire- to investigate how students regulate their learning strategies in more realistic learning situations. Four scales were created: (1) Basic learning self-regulation strategies; (2) Visual elaboration and summarizing strategies; (3) Deep information processing strategies; and (4) Social learning self-regulation strategies. A total of 601 higher education students formed the sample. We analyzed, first, the internal validity of the questionnaire. Three structural models were tested: (M1) mono-factor; (M2) scales correlate among them freely, and (M3) the scales are indicators of a general construct. The latter model showed a slight better fit. Additionally, a path analysis was carried out to study the degree in which the use of the Deep learning strategies depends on personal factors and is associated to performance. It was found that the use depends directly and positively on learning goal orientation, on the self-messages defining the self-regulation style of emotion and motivation focused on learning, and on effort. Besides, these two last variables convey the effect of self-efficacy that, at the same time, affects effort. Academic performance, depends positively on effort but negatively to the use of deep learning strategies. It is hypothesized this negative relationship is due to the method of measurement of academic performance.

Predictors of K-12 Teachers’ Instructional Strategies with ICTs

The goal of this study is to identify the relationship between K-12 teachers’ self-efficacy beliefs, classroom goal structure and use of instructional strategies. The study also aims to determine if there is variance in the relationship between these constructs for primary versus secondary school teachers. Data collection involved completion of a self-report survey by 810 primary and secondary school teachers in Thailand. Results revealed that personal and ICT teaching self-efficacy directly predicted both mastery and performance classroom goal structures for K-12 teachers. Mastery classroom goal structure predicted deep learning. Performance classroom goal structure predicted surface learning. ICT teaching self-efficacy was the strongest predictor of teachers’ instructional strategies with ICTs. Results also revealed that primary teachers’ performance classroom goal structure was positively associated with the use of deep-learning strategies with ICTs and student-centered learning with ICTs. In contrast, for secondary teachers, mastery classroom goal structure was positively associated with student-centered learning with ICTs, whereas both mastery and performance classroom goal structures were positively associated with the use of deep-learning strategies with ICTs. Results of this study suggest that classroom goal structure may be different for ICT classrooms than for regular classrooms. Implications relate to the need to help teachers design ICT activities that reflect both performance and mastery classroom goal structures.

Real-time shadow detection using multi-channel binarization and noise removal

High-quality automatic shadow detection remains a challenging problem in image processing and computer vision. Existing techniques for shadow detection typically make use of deep learning strategies to obtain accurate shadow detection results, at the cost of demanding high processing time, making their use unsuitable for augmented reality and robotic applications. In this paper, we propose a novel approach to perform high-quality shadow detection in real time. To do so, we convert an input image into different color spaces to perform multi-channel binarization and detect different shadow regions in the image. Then, a filtering algorithm is proposed to remove the noisy false-positive shadow regions on the basis of their sizes. Experimental results evaluated in two different datasets show that the proposed approach may run entirely on the GPU, requiring only $$\approx$$ 13 ms to detect shadows in an image with $$3840 \times 2160$$ (4k) resolution. That makes our approach about 1.8 (66$$\times$$) to 4.6 (37,284$$\times$$) orders of magnitude faster than related work for 4k resolution images, at the cost of only $$\approx$$ 5% of accuracy loss compared to the best results achieved for each dataset.

Motivational Profiles of Adult Learners

This study investigated profiles of autonomous and controlled motivation and their effects in a sample of 188 adult learners from two Portuguese urban areas. Using a person-centered approach, results of cluster analysis and multivariate analysis of covariance revealed four motivational groups with different effects in self-efficacy, engagement, and learning. The study showed that groups of learners who have high autonomous motivation in the beginning of a course score higher in self-efficacy and later on in behavioral engagement and use of deep-learning strategies, whereas those who have controlled motivation alone or low levels of both types of motivation have worse results. Additionally, the study showed motivational differences according to adult learners’ gender, educational level, and occupational status. The influence of the Portuguese adult education system on the results and the implications of the study for the practice of adult education are also discussed.

An Exploratory Study of Studentss Lived Experiences in a Blended-Flipped Writing Class

During the past few decades blended instructional approaches have aimed to provide learners with an enhanced learning experience. Supported by a flipped approach to teaching, blended instruction provides an opportunity for an effective utilization of classroom time. This article reports the findings of a mixed-methods study which involved 57 students enrolled in Level 3 academic writing course offered as part of the General Foundation Program (GFP) in a non-public higher education institution in Oman. The purpose of the study was to explore the lived experiences of EFL (English as Foreign Language) learners in a blended flipped writing course with specific focus on their perceptions of the benefits and challenges of being instructed in this method. The study informs both EFL educators and learners, as well as other stakeholders, about this method of teaching and triggers further discussions of its potential and limitations in EFL instruction. This study utilized students’ self-report questionnaire, teacher observation, and focus group interviews as data collection methods; and value thematic qualitative data coding and descriptive quantitative data analysis as data analysis procedures. The study revealed that despite the challenges of increased workload and technology-related difficulties, participants in the blended-flipped academic writing class experienced increased involvement, development and use of deep learning strategies, and maximization of learning through constant collaboration. The study recommends that further experimental studies are conducted to explore the potential of blended-flipped instruction in teaching other English language skills and the correlation between the learners’ perceptions of this instructional model and their academic performance.

Self-Regulation in Higher Education: Students’ Motivational, Regulational and Learning Strategies, and Their Relationships to Study Success

Abstract This study investigates how in the self-regulation of learning (SRL; Pintrich 2000; Zimmerman, 2000), the motivational and affective factors are related to regulation strategies of behaviour and context, and learning strategies - and identifies different profiles in SRL. The study also aims to explore which factors of SRL are related to study success and study progress during master degree studies. The data consist of undergraduate students’ (N = 1248) responses to IQ Learn self-report questionnaires, and of data (n = 229) retrieved from the university ’ s study register. The results revealed that the sub-processes of SRL: motivational and affective components, regulation strategies and learning strategies are systematically related with each other. In addition, motivational and affective factors, especially Intrinsic motivation predicted the use of strategies regulating behaviour and context and the use of learning strategies. Study success correlated slightly positively with accumulation of credits indicating that students with better grades proceed efficiently in their studies. Yet, accumulation of credits was evidenced to relate slightly and negatively with expectancy components of SRL and the use of deep learning strategies. Finally, three student profiles in SRL were encountered: (1) Aiming high with insufficient SRL, (2) Excellent in SRL, and (3) Distressed performers. Educational implications and the needs for future research are discussed.

The Combined Effects of Classroom Teaching and Learning Strategy Use on Students’ Chemistry Self-Efficacy

For students to be successful in school chemistry, a strong sense of self-efficacy is essential. Chemistry self-efficacy can be defined as students’ beliefs about the extent to which they are capable of performing specific chemistry tasks. According to Bandura (Psychol. Rev. 84:191–215, 1977), students acquire information about their level of self-efficacy from four sources: performance accomplishments, vicarious experiences, verbal persuasion, and physiological states. No published studies have investigated how instructional strategies in chemistry lessons can provide students with positive experiences with these four sources of self-efficacy information and how the instructional strategies promote students’ chemistry self-efficacy. In this study, questionnaire items were constructed to measure student perceptions about instructional strategies, termed efficacy-enhancing teaching, which can provide positive experiences with the four sources of self-efficacy information. Structural equation modeling was then applied to test a hypothesized mediation model, positing that efficacy-enhancing teaching positively affects students’ chemistry self-efficacy through their use of deep learning strategies such as metacognitive control strategies. A total of 590 chemistry students at nine secondary schools in Hong Kong participated in the survey. The mediation model provided a good fit to the student data. Efficacy-enhancing teaching had a direct effect on students’ chemistry self-efficacy. Efficacy-enhancing teaching also directly affected students’ use of deep learning strategies, which in turn affected students’ chemistry self-efficacy. The implications of these findings for developing secondary school students’ chemistry self-efficacy are discussed.

Are two achievement goals better than one? Filipino students' achievement goals, deep learning strategies and affect

The multiple-goals perspective assumes that adopting both mastery and performance goals is better for students. But there has been very little evidence for the multiple-goals perspective in Asian students. A cluster analysis was conducted to identify groups of Filipino students who endorse different goal combinations, revealing four groups comprised of those who adopted (a) predominantly performance goals, (b) predominantly mastery goals, (c) multiple goals or (d) low levels of both goals. Comparisons of cognitive and affective learning variables indicated that reported use of deep learning strategies, enjoyment of learning and less anxiety were associated with the multiple goals group. Although the study did not assess actual learning outcomes, the study shows results consistent with the multiple-goals perspective, as it shows enhanced use of cognitive strategies and more positive affective experiences in the learning processes, adding to the scant evidence on multiple-goals perspective among Asian students.

The effects of classroom teaching on students' self-efficacy for personal development

The personal development of students is an essential component of school guidance and counselling programmes, but no published research on guidance and counselling has investigated the effects of regular classroom teaching on students' self-efficacy for personal development. In this study, questionnaire items were constructed to measure classroom teaching, student self-efficacy for personal development and student use of deep learning strategies. Data were collected from 16,208 secondary school students in Hong Kong. Using structural equation modelling, regular classroom teaching was found to have a direct effect on personal development self-efficacy as well as an indirect effect through student use of deep learning strategies. Implications of these findings for implementing a whole-school approach to guidance and counselling are discussed.

An Investigation Of Ethnic Differences In The Motivation And Strategies For Learning Of Students In Desegregated South African Schools

No significant ethnic group differences were found in the academic motivation or learning strategies of more than 200 White and 200 Black South African students from two recently integrated secondary schools. Both groups considered working hard and having interest in school tasks to be more important than competition or affiliation. The former items tended to be the best predictors in both groups of the use of deep learning strategies, which are known to be prerequisites for high-quality learning outcomes. The motivation scales tended to be better predictors of learning strategies for the Black students.