Deep Learning Outcomes Research Articles

Learning With Board Games: Interplay of Context and Students’ Prior Knowledge

Board games can support engagement, deep information processing and learning outcomes for university students. The instructional context, notably the way an activity is introduced to students, can influence students’ behavior and learning outcomes. Yet, despite its potential important influence on in-game behavior and learning, the impact of the instructional context used to introduce a game-based learning activity has not received much attention. In two studies, we examined how framing a game-based learning activity with a serious or a non-serious context can impact university students’ behavior and learning outcomes. In the first study, 44 graduate students played the same word association game that was either introduced as a game or as a study activity. Results showed that the serious context increased recall as compared to the gaming context. In the second study, involving 87 participants playing the same game, results showed that the gaming context and prior gaming experience also reduced students’ recall. Overall, grouped analysis showed that the gaming context led to poorer learning than the serious context for recognition and recall measures, even when considering prior knowledge, in-game behavior, and prior gaming experience. Theoretical and practical implications for educators are discussed.

Acoustic conditions in innovative learning environments

New generation or innovative learning environments are being developed in many OECD coun-tries, including Australia and New Zealand. These spaces are often criticized as being 'noisy', dis-tracting, unattenable, and yet, there has been a link between deep learning outcomes and the new generation environments. Education spaces primarily revolve around speaking and listening activities. Although new space typologies have been, and continue to be, developed, the acoustic standards applied to them are often the same as those applied to traditional cellular classroom spaces. This paper presents the results of systematic acoustic testing undertaken within numerous physical spaces for traditional and new generation learning environments in Australia and New Zealand, to compare the differences in performance between traditional cellular classrooms and open plan spaces, and summarise any trends evident in the information. It also acknowledges the differences between traditional cellular spaces and new generation environments as determined from the results.

DeepGAN: Utilizing generative adversarial networks for improved deep learning

In the realm of deep learning, Generative Adversarial Networks (GANs) have emerged as a topic of significant interest for their potential to enhance model performance and enable effective data augmentation. This paper addresses the existing challenges in synthesizing high-quality data and harnessing the capabilities of GANs for improved deep learning outcomes. Unlike traditional approaches that heavily rely on manually engineered data augmentation techniques, our work introduces a novel framework that leverages DeepGANs to autonomously generate diverse and high-fidelity data. Our experiments encompass a diverse spectrum of datasets, including images, text, and time series data. In the context of image classification tasks, we conduct experiments on the widely recognized CIFAR-10 dataset, which consists of 50,000 image samples. Our results demonstrate the remarkable efficacy of DeepGANs in enhancing model performance across various data domains. Notably, in image classification using the CIFAR-10 dataset, our innovative approach achieves an impressive accuracy of 97.2%. This represents a substantial advancement beyond conventional CNN models, underscoring the profound impact of DeepGANs in the realm of deep learning. In summary, this research sheds light on DeepGANs as a fundamental component in the pursuit of enhanced deep learning performance. Our framework not only overcomes existing limitations but also heralds a new era of data augmentation, with generative adversarial networks leading the way. The attainment of an accuracy rate of 97.2% on CIFAR-10 serves as a compelling testament to the transformative potential of DeepGANs, solidifying their pivotal role in the future of deep learning. This promises the development of more robust, adaptive, and accurate models across a myriad of applications, marking a significant contribution to the field.

Design and implementation effects of SPOC-based blended teaching from the perspective of deep teaching: a case study of EFL students

ABSTRACT While blended teaching approaches are becoming more common in some higher educational institutes, their consistency in aligning with higher-level learning objectives varies The current study aims to explore the design and effectiveness of Small Private Online Course (SPOC)-based blended teaching in fostering deep learning among English as a Foreign Language students in Higher Vocational Colleges in Guangdong, China. We develop a systematically designed SPOC-based blended teaching approach tailored for deep learning promotion and compare it with a non-systematic SPOC-based design. The study then implements the SPOC-based blended teaching approach to promote deep learning and collects relevant data for analysis. Our results show significant improvements in students’ knowledge acquisition, skill development, and overall satisfaction compared to the non-systematic approach. This study provides valuable insights into deep blended teaching and underscores the necessity of a systematic shift in teaching methodologies through SPOC-based blended learning to optimise deep learning outcomes.

Multi-channel microseismic signals classification with convolutional neural networks

Identifying and classifying microseismic signals is essential to warn of mines’ dangers. Deep learning has replaced traditional methods, but labor-intensive manual identification and varying deep learning outcomes pose challenges. This paper proposes a transfer learning-based convolutional neural network (CNN) method called microseismic signals-convolutional neural network (MS-CNN) to automatically recognize and classify microseismic events and blasts. The model was instructed on a limited sample of data to obtain an optimal weight model for microseismic waveform recognition and classification. A comparative analysis was performed with an existing CNN model and classical image classification models such as AlexNet, GoogLeNet, and ResNet50. The outcomes demonstrate that the MS-CNN model achieved the best recognition and classification effect (99.6% accuracy) in the shortest time (0.31 s to identify 277 images in the test set). Thus, the MS-CNN model can efficiently recognize and classify microseismic events and blasts in practical engineering applications, improving the recognition timeliness of microseismic signals and further enhancing the accuracy of event classification.

A Quantitative Study on Undergraduate Students’ Deeper Learning in Peer-Feedback and Self-Feedback Tasks in an English as a Foreign Language Writing Course

Deeper learning competencies, such as critical thinking, communication, collaboration, and self-management, are receiving increasing attention in education. In writing instruction, feedback-providing tasks have the potential for promoting deeper learning. This study develops and applies a self-report scale to explore the level of deeper learning among undergraduate students in China in providing peer feedback and self-feedback in an English as a Foreign Language (EFL) writing course. The scale includes three dimensions: involvement, use of deeper learning strategies, and deeper learning outcomes. The research found a generally moderate level of deeper learning in various feedback-giving tasks. Significant and positive correlations were found between the three dimensions of deeper learning as well as between deeper learning, anticipation toward feedback-giving tasks and overall course evaluation. At the same time, it was found that students may be more active in deeper learning from the perspectives of their use of deeper learning strategies and the perceived learning outcomes than from the perspectives of participation in discussion as well as time and effort spent on the tasks.

Exploring the effects of digital technology on deep learning: a meta-analysis

AbstractThe impact of digital technology on learning outcomes, specifically deep learning, has been a subject of considerable debate and scrutiny in educational settings. This study aims to provide clarity by conducting a meta-analysis of empirical publications that examine students' deep learning outcomes in relation to digital technology. A comprehensive search of databases and a thorough literature review yielded 60 high-quality, peer-reviewed journal articles that met the inclusion criteria. Using Review Manager 5.4.1 software, a meta-analysis was conducted to assess the overall effectiveness of digital technology. The calculated effect size indicates a positive influence of digital technology on students' deep learning outcomes. Furthermore, a moderator variable analysis revealed several significant findings: 1. Different categories of digital technology tools have a favorable impact on deep learning outcomes; 2. The duration of digital technology treatment does not significantly affect deep learning outcomes; 3. Digital technology demonstrates a highly positive influence on deep learning within the humanities and social sciences disciplines; 4. Combining online and offline utilization of digital technology in education leads to a substantially greater enhancement in deep learning compared to relying solely on online methods; 5. The effectiveness of digital technology on deep learning is enhanced when accompanied by appropriate instructional guidance; 6. Utilizing digital technology in a systematic manner produces different outcomes compared to fragmented approaches, highlighting the importance of a cohesive implementation; 7. Integrating digital technology with collaborative learning has a more pronounced effect on deep learning compared to independent learning. These findings contribute to our understanding of the impact of digital technology on deep learning outcomes and underscore the importance of thoughtful integration and instructional support in educational contexts.

Physics-Guided Deep Generative Model for New Ligand Discovery.

Structure-based drug discovery aims to identify small molecules that can attach to a specific target protein and change its functionality. Recently, deep learning has shown great promise in generating drug-like molecules with specific biochemical features and conditioned with structural features. However, they usually fail to incorporate an essential factor: the underlying physics which guides molecular formation and binding in real-world scenarios. In this work, we describe a physics-guided deep generative model for new ligand discovery, conditioned not only on the binding site but also on physics-based features that describe the binding mechanism between a receptor and a ligand. The proposed hybrid model has been tested on large protein-ligand complexes and small host-guest systems. Using the top-N methodology, on average more than 75% of the generated structures by our hybrid model were stronger binders than the original reference ligand. All of them had higher ΔGbind (affinity) values than the ones generated by the previous state-of-the-art method by an average margin of 1.88 kcal/mol. The visualization of the top-5 ligands generated by the proposed physics-guided model and the reference deep learning model demonstrate more feasible conformations and orientations by the former. The future directions include training and testing the hybrid model on larger datasets, adding more relevant physics-based features, and interpreting the deep learning outcomes from biophysical perspectives.

An Exploration of Students' Engagement in EFL Asynchronous Class

Students’ engagement is an ongoing concern for educators because of its positive association with deep learning, educational purpose, and educational outcomes. The purpose of this study is to reveal the students’ engagement and to explore the factors of students’ engagement in EFL asynchronous classes. This research was qualitative design with a case study approach. The participants of this study were 35 ten-grades students in one of the senior high schools in Majalengka. Survey and interview were used to gain the data from the participants. The survey was conducted to 35 ten-graders which mainly focused on their engagement during the asynchronous class as well as their reasons of being to or not engage with the lessons. Meanwhile, factors influenced their engagement in the asynchronous class were elaborated using the interview. The result showed that students gave positive responses in students’ engagement on cognitive and behavioral engagement, most of the students both high and middle achievers always follow the teacher instructions and sometimes give opinions/ideas. The most prominent factors that influence students’ engagement is teacher factors, among them are interesting teaching method and teachers’ characteristic This study suggests that teachers need a variety of methods in which to use online platforms and conduct more creative activities in order to get the best learning outcomes.Keywords: Asynchronous Learning, EFL Class, Student Engagement.

Perceived Connection to Instructor and Instructor Passion as Predictors of Transformative Experiences in Science

Transformative experience represents a deep-engagement construct and refers to experiences in which students use science content to see and experience the world in meaningful new ways outside the classroom. Such experiences are associated with deep learning (e.g., transfer) and engagement (e.g., academic and career choice) outcomes. However, research on individual factors predictive of transformative experience is limited. The current study investigated university geoscience students’ perceptions of a connection to their instructor(s) and perceptions of their instructors’ passion for the content as predictors of transformative experience. Controlling for students’ initial interest and self-efficacy in the domain, we found both these factors to be predictive of transformative experience. These results suggest science teachers may be able to support engagement in transformative experiences by connecting with students and expressing a passion for the content.

Learning from text, video, or subtitles: A comparative analysis

The present study investigated the influence of media (text, video, or subtitled video) on students' learning outcomes. Past studies have raised concerns about the effectiveness of learning from online videos over content-equivalent texts. Moreover, subtitled videos place additional demands on learning. Two-hundred and forty-seven undergraduate students were randomly assigned to a text, video, or subtitled-video condition, in a pretest, posttest, and delayed posttest design. The topic assigned was stem cells. Literal, inferential, and transfer questions were used to assess comprehension and learning outcomes. Results from the study confirmed the substantial equivalence of all conditions in immediate comprehension. Conversely, results confirmed the disadvantage of subtitled videos for deep learning outcomes.

Is there deep learning on Mars? STEAM education in an inquiry-based out-of-school setting

ABSTRACT Learning intervention based on a “Mars and Space” exhibition was designed according to STEAM-education (Science, Technology, Engineering, Art and Mathematics) principles and practices in order to bridge the gap between formal and informal learning. The cognitive learning of 12-year-old students in Finland (N = 306) showed a sustained level for a six months period. The results of this study provided evidence that situational motivation was enhanced by interest in school science in the STEAM science exhibition context. This led to better cognitive learning results in the post-knowledge test. Thus, interest and situational motivation were the first steps, and the superficial situational motivation seemed to successfully change into content-based intrinsic motivation with longer-lasting, deep learning outcomes. STEAM intervention apparently produced long-term learning, and this exhibition learning setting is shown to provide an appropriate platform to reach the deeper layers to successfully retain knowledge. Boys’ scores rose even in the delayed test. Using structural equation modelling (SEM) to assess the effects on individual, motivational and situational interest in learning, situation motivation is shown to work as a catalyst and acting as a catalyst and also acted as a stepping stone for intrinsic motivation as part of relative autonomy (RAI) and a deep learning strategy.

Utility of a self-regulated learning microanalysis for assessing learning during professional development

A key process in the discipline of science is argumentation, which has recently been receiving more attention in K-12 classrooms. The purpose of this multiple case study was to concurrently examine self-regulated learning processes (SRL) and learning outcomes of three teachers in a professional development (PD) on argumentation in science to assess the relationship between types of learning processes employed by teachers and corresponding learning outcomes. Three teacher cases were selected based on variations in their self-efficacy for the PD, the type of schools in which they taught, and their experiences with argumentation in science. Teachers who were able to use SRL strategies had deeper learning outcomes and created lesson plans that engaged students in higher levels of argumentation. It was found that the SRL microanalysis was useful in eliciting learning processes for learners and provided opportunities for learners to be more aware of their SRL practices. Implications for this study include ways to measure learning processes and learning outcomes concurrently to provide more effective PD instruction.

Dropout vs. batch normalization: an empirical study of their impact to deep learning

Overfitting and long training time are two fundamental challenges in multilayered neural network learning and deep learning in particular. Dropout and batch normalization are two well-recognized approaches to tackle these challenges. While both approaches share overlapping design principles, numerous research results have shown that they have unique strengths to improve deep learning. Many tools simplify these two approaches as a simple function call, allowing flexible stacking to form deep learning architectures. Although their usage guidelines are available, unfortunately no well-defined set of rules or comprehensive studies to investigate them concerning data input, network configurations, learning efficiency, and accuracy. It is not clear when users should consider using dropout and/or batch normalization, and how they should be combined (or used alternatively) to achieve optimized deep learning outcomes. In this paper we conduct an empirical study to investigate the effect of dropout and batch normalization on training deep learning models. We use multilayered dense neural networks and convolutional neural networks (CNN) as the deep learning models, and mix dropout and batch normalization to design different architectures and subsequently observe their performance in terms of training and test CPU time, number of parameters in the model (as a proxy for model size), and classification accuracy. The interplay between network structures, dropout, and batch normalization, allow us to conclude when and how dropout and batch normalization should be considered in deep learning. The empirical study quantified the increase in training time when dropout and batch normalization are used, as well as the increase in prediction time (important for constrained environments, such as smartphones and low-powered IoT devices). It showed that a non-adaptive optimizer (e.g. SGD) can outperform adaptive optimizers, but only at the cost of a significant amount of training times to perform hyperparameter tuning, while an adaptive optimizer (e.g. RMSProp) performs well without much tuning. Finally, it showed that dropout and batch normalization should be used in CNNs only with caution and experimentation (when in doubt and short on time to experiment, use only batch normalization).

Student Engagement and Deep Learning in the First-Year International Relations Classroom: Simulating a UN Security Council Debate on the Syrian Crisis

This article evaluates a “real-time” simulation where students role-play a United Nations Security Council negotiation over humanitarian intervention in Syria. This simulation is undertaken in a large introductory International Relations (IR) subject. The article argues that in order to achieve deep learning outcomes across the diverse, contemporary cohort of first-year university students, active learning approaches need to be employed that engage differing learning styles and preferences. Deep learning is assessed across the conceptual and metacognitive knowledge domains with two indicators: (1) students’ understanding of IR concepts by applying them beyond the parameters of the Syria case and (2) students’ critical reflection on their moral reasoning elicited by the task. We evaluate 820 students across six cohorts and 21 iterations of the simulation during 2016 and 2017 with a survey instrument and formal reflection assignment. The article finds that the simulation is highly effective at consistently engaging the majority of students’ interest and motivation, while illustrating the acute and sometimes tragic tension between moral and political reasoning in IR. We found that disrupting student’s cognitive structures regarding human rights and justice stimulated not only deeper conceptual understanding but also emotional reactions that were the catalyst for metacognitive reflection.

GENDER, VOICE, AND LEARNING IN ONLINE COURSE ENVIRONMENTS

A self-selected sample of 109 online students at a midwestern regional university was surveyed and asked to compare expression of voice, control over learning, and perceived deep learning outcomes in face-to-face versus online course environments. We found that females experience greater perceived deep learning in online than in face-to-face courses, and that expression of voice appears to contribute to this outcome. This effect did not occur for male students. We also found that professor support and, to a lesser extent, control over one's learning each had positive relationships with perceived deep learning in both course environments. Concern for the feelings of other students did not have a negative impact on voice as was originally hypothesized.

3D Randomized Connection Network with Graph-based Label Inference.

In this paper, a novel 3D deep learning network is proposed for brain MR image segmentation with randomized connection, which can decrease the dependency between layers and increase the network capacity. The convolutional LSTM and 3D convolution are employed as network units to capture the long-term and short-term 3D properties respectively. To assemble these two kinds of spatial-temporal information and refine the deep learning outcomes, we further introduce an efficient graph-based node selection and label inference method. Experiments have been carried out on two publicly available databases and results demonstrate that the proposed method can obtain competitive performances as compared with other state-of-the-art methods.

Manipulating cognitive engagement in preparation-to-collaborate tasks and the effects on learning

While collaborating with a peer can be highly beneficial for learning, more work is needed to understand how instructional activities in collaborative contexts should be designed so as to maximize learning outcomes. To address this, we investigated the impact of different types of preparatory and cognitively engaging tasks on learning from collaborating, using a 2 × 2 experimental study conducted in situ in four introductory psychology classes. We compared individual preparation versus no-preparation and “active” versus “constructive” tasks. A dyadic multilevel analysis showed that preparation prior to collaborating led to better deep learning outcomes, but that the type of preparation did not have a significant effect. We include an exploratory analysis of student dialogues during collaboration to further interpret our findings. We propose that a cognitively engaging preparation phase may lead to better learning because it encourages students to collaborate constructively even when the type of task does not elicit such engagement.

Presence and learning in a community of inquiry

The community of inquiry (CoI) framework suggests social presence, teaching presence, and cognitive presence are essential elements to foster successful educational experiences in computer-mediated distance learning environments. Although thousands of CoI-based articles have been published, those critical of the framework and related research suggest a lack of empirical evidence to support the framework’s central claim that a CoI fostering learners’ social, teaching, and cognitive presence leads to deep and meaningful learning outcomes. The current study conducted with 51 graduate students in five distance education courses compared student responses to a CoI perception survey with three instructor-assessed learning achievement measures. No relationship was found between the students’ perceptions of their participation within the CoI and any of the three instructor-assessed learning achievement measures, suggesting the value of the CoI framework as an educational process model remains challenged.

The High Quality Business School Academic Teacher of the 21st Century – Polish Students’ Perspective

The literature shows that the success and competence of future managers depend on the quality of their academic teachers. Moreover high quality study requires high quality lecturing/teaching that creates an environment in which deep learning outcomes are made possible for students. The aim was to identify the characteristics of the academic teacher working at business schools, according to the expectations of Polish students from a 21st century perspective. A qualitative survey design was used in the form of a letter questionnaire. 144 second-year bachelor students of Gdansk University from the Faculty of Management were asked to list a maximum of five, most preferred characteristics, and to comment their answers. Finally, 109 students participated in the study, and 471 characteristics were proposed, analysed and put into five categories: tangibles (T), reliability (Rel), responsiveness (Res), assurance (A) and empathy (E). Content analysis and Pareto-Lorenzo analysis was used and the most preferable characteristics were identified. Conclusions, proposals and recommendations were presented. The academic teacher has to be well prepared and teach in an interesting, innovative way with a use of modern techniques and methods. Very important is to apply not only the lecture-style methods, but also on-the job teaching, project-based teaching, team work-based teaching, action teaching, experiential teaching, small groups teaching, case studies, simulations, e-teaching, and even volunteering teaching. Not without the significance are coaching and mentoring and the features referring to the style of teaching, like charisma, creativity, passion, and engagement, which characterise good managers and business leaders.