• Electronics and Telecommunications Trends
• /
• v.34 no.6
• /
• pp.87-99
• /
• 2019

Reinforcement learning is a type of machine learning paradigm that forces agents to repeat the observation-action-reward process to assess and predict the values of possible future action sequences. This allows the agents to incrementally reinforce the desired behavior for a given observation. Thanks to the recent advancements of deep learning, reinforcement learning has evolved into deep reinforcement learning that introduces promising results in various control and optimization domains, such as games, robotics, autonomous vehicles, computing, industrial control, and so on. In addition to this trend, a number of programming libraries have been developed for importing deep reinforcement learning into a variety of applications. In this article, we briefly review and summarize 10 representative deep reinforcement learning libraries and compare them from a development project perspective.

• International journal of advanced smart convergence
• /
• v.12 no.4
• /
• pp.164-170
• /
• 2023

With the recent advances in AI (artificial intelligence) and HPC (high-performance computing) technologies, deep learning is proliferated in various domains of the 4th industrial revolution. As the workload volume of deep learning increasingly grows, analyzing the memory reference characteristics becomes important. In this article, we analyze the memory reference traces of deep learning workloads in comparison with traditional workloads specially focusing on read and write operations. Based on our analysis, we observe some unique characteristics of deep learning memory references that are quite different from traditional workloads. First, when comparing instruction and data references, instruction reference accounts for a little portion in deep learning workloads. Second, when comparing read and write, write reference accounts for a majority of memory references, which is also different from traditional workloads. Third, although write references are dominant, it exhibits low reference skewness compared to traditional workloads. Specifically, the skew factor of write references is small compared to traditional workloads. We expect that the analysis performed in this article will be helpful in efficiently designing memory management systems for deep learning workloads.

• Journal of Gifted/Talented Education
• /
• v.26 no.1
• /
• pp.77-99
• /
• 2016

The purpose of this study was to examine the effect of GI-STEAM program on leadership, creative personality, and learning flow of elementary Gifted Students. GI-STEAM program was the convergence model of Group Investigation that belongs to Co-learning and STEAM framework of learning criterion. The participants were 16 gifted students in a Korean elementary school located in Gyeong-gi province. The experimental design was one group pretest-posttest design. After a pretest on leadership, creative personality, and learning flow was conducted, classes were carried out as GI-STEAM program for the gifted student and a post-test was conducted. The study results of the class that was conducted twelve times for two weeks are as follows. First, Individual area of leadership is meaningfully developed in statistics after GI-STEAM program. The sub-domains of leadership, such as the communication, organization management, society commitment and teamwork showed a statistically significant improvement. Second, the domain of creative personality didn't show meaningful difference after GI-STEAM program. However, the aesthetic in the sub-domains of the creative personality showed a statistically significant improvement. Third, learning flow was meaningfully developed in statistics after GI-STEAM program. The sub-domains of the leadership, such as the balance between challenge and ability, integration with behavior and consciousness, concrete feedback and Autotelic experience showed a statistically significant improvement. In conclusion, GI-STEAM is an effective program for improving ability of communication, aesthetic sensibility, which are core competency of 'creative-convergence' gifted students. For this reason, it is highly considered that various programs applying GI-STEAM should be developed.

• Journal of The Korean Association For Science Education
• /
• v.22 no.3
• /
• pp.540-549
• /
• 2002

The 7th national curriculum is focused on breeding an independent and creative Korean who will lead the age of globalism and information in the 21st century. It is necessary to improve the existing assessment methods in order to develop higher thinking abilities such as creativity and problem-solving skill. Although teachers have been aware of this necessity, they have realized that it is difficult to improve the current assessment methods. In this study, we selected some assessment domains on science learning with literature reviews and case analysis. In addition, we calculated the degree of its importance by the use of Analytic Hierarchy Process(AHP). We suggest a direction for improving the present assessment domains on science learning on the basis of the research. Inquiry, cognitive, creative, and affective domain among assessment domains seemed to be listed in order of importance. Moreover, problem-identifying, hypothesizing, and inquiry-planning appeared to be the highest in the degree of importance among sub categories. Considering the results of this study, the current school assessment system which is focused on cognitive domain should be improved.

• English Language & Literature Teaching
• /
• v.9 no.2
• /
• pp.263-291
• /
• 2003

This paper is to help the students in elementary schools develop and improve their English listening and speaking skills by presenting effective teaching and learning techniques using puppetry. It is absolutely obvious that listening and speaking are very important skills for most EFL students. Using puppets in the classroom is a creative English teaching technique which can involve authentic, communicative language situations. Moreover, puppets appeal to children and can aid in lowering affective filters thereby creating a more comfortable learning environment. The study clearly showed that using puppets is feasible and enjoyable in elementary English classes. However, caution must be exercised in drawing and generalizing conclusions from this experience. The results of the experiment are as follows: First, using puppetry in the English class was found to have positive influence on students' affective domains (interst, attitude). Second, using puppets in the English classes was found to be efficient for improving students' English listening and speaking skills. Third, appropriate materials should be selected and well thought-out plans should be made to be successful English class using puppetry. Perhaps the most interesting line of future research is to use qualitative research to examine the effect of this technique on the teacher variable. Further research is recommended, especially on using puppetry for speaking proficiency and creating affectively comfortable learning atmospheres.

• Communications of Mathematical Education
• /
• v.32 no.1
• /
• pp.97-111
• /
• 2018

This study is to establish the domains and the standards of instructional evaluation on the teacher knowledge dealing with the knowledge of 'teaching and learning methods and assessment'. Especially, in this study, the instruction assessment standards are developed focused on the six types of mathematics competencies such as problem solving, communication, reasoning, creativity and collaboration, information and handling, attitude and practice which were emphasized in the mathematical curriculum revised in 2015. By the result, seventh evaluation domains such as an instruction involving problem-solving activity, an instruction involving reasoning activity, instruction involving communication activity, instruction on information and handling activity, instruction involving learners' achievement level and attitude, instruction involving the development of assessment method and tool, instruction applying on assessment result were new established. According to those domains, the 19 instructional evaluation standards were developed totally. This study is limited to consider the domain of 'teaching and learning methods and assessment' among the domains of teacher knowledge, while dealing with the elements of mathematics competencies in the standards. However, instructional evaluation standards reflecting these competencies should be developed in the other diverse domains of teacher knowledge.

• Journal of Broadcast Engineering
• /
• v.20 no.5
• /
• pp.676-686
• /
• 2015

A 4D light field image is represented in traditional 2D spatial domain and additional 2D angular domain. The 4D light field has a resolution limitation both in spatial and angular domains since 4D signals are captured by 2D CMOS sensor with limited resolution. In this paper, we propose a dictionary learning-based superresolution algorithm in 4D light field domain to overcome the resolution limitation. The proposed algorithm performs dictionary learning using a large number of extracted 4D light field patches. Then, a high resolution light field image is reconstructed from a low resolution input using the learned dictionary. In this paper, we reconstruct a 4D light field image to have double resolution both in spatial and angular domains. Experimental result shows that the proposed method outperforms the traditional method for the test images captured by a commercial light field camera, i.e. Lytro.

• Journal of The Korean Association For Science Education
• /
• v.18 no.4
• /
• pp.539-544
• /
• 1998

The consistency and balance of objectives by objective domains in units, sub-units and instructional units were investigated. The 6th elementary biology curricular teaching guidebooks were analysed. Domains of objectives are cognitive, inquiry process, instrumental skill, creative, affective and STS. Cognitive objectives were most dominant in all units, sub-units and instructional units. But no objective for creative domain were suggested. In unit and sub-unit, proportions of objectives were cognitive, inquiry process, affective, instrumental skill and STS domains in order. Objectives for cognitive and inquiry process domains were more than others in instructional units, Except cognitive and inquiry process domains, objectives for the others were not consistent in all units, sub-units and instructional units. Especially, the percentages of objectives for affective domain decreased in units, sub-units and instructional units orderly. These resulted from teaching objective domains categorized formally, Thus, it is necessary to develope curriculum and textbook to be consistent and balanced with objective domains and reflect upon the characteristics of them.

• 대한공업교육학회지
• /
• v.34 no.2
• /
• pp.205-229
• /
• 2009

The purpose of this study is to verify the effect of forming groups according to the brain hemisphere preference on the cooperative problem solving learning achievement in the middle school technology. The subjects of this study were 95 second grade boy students of a middle school in Daejeon and the measurement instrument of the left and right hemisphere preference is the Brain preference Indicator(BPI) which had been developed by Torrance et al(1977) and was adjusted by Ko, Younghee(1991). The academic achievement was analyzed on cognitive, psychomotor and affective domains. Derived results from this research are stated below: First, making groups according that the brain preference is more similar was more effective than making groups according to the high familiarity and the similarity of performance in the academic achievement of psychomotor and affective domains. Second, making groups according that the brain preference is more similar was more effective than making groups according that the brain preference is more diffrent for the academic achievement of affective domains on the cooperative problem solving learning in technology. Third, the academic achievement score of the right hemisphere preference group is higher than the score of the population in three domains. Also, the academic achievement score of the right hemisphere preference group is higher than the score of the left hemisphere preference group.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• v.14 no.3
• /
• pp.209-215
• /
• 2014

Sequence tagging is the task of predicting frame-wise labels for a given input sequence and has important applications to diverse domains. Conventional methods such as maximum likelihood (ML) learning matches global features in empirical and model distributions, rather than local features, which directly translates into frame-wise prediction errors. Recent probabilistic sequence models such as conditional random fields (CRFs) have achieved great success in a variety of situations. In this paper, we introduce a novel discriminative CRF learning algorithm to minimize local feature mismatches. Unlike overall data fitting originating from global feature matching in ML learning, our approach reduces the total error over all frames in a sequence. We also provide an efficient gradient-based learning method via gradient forward-backward recursion, which requires the same computational complexity as ML learning. For several real-world sequence tagging problems, we empirically demonstrate that the proposed learning algorithm achieves significantly more accurate prediction performance than standard estimators.