Pub Date : 2021-01-01DOI: 10.6688/JISE.202101_37(1).0015
J. Kuo, Hao-Ting Chung, Ping-Feng Wang, Baiying Lei
The deferral of graduation rate in Taiwan's universities is estimated 16%, which will affect the scheduling of school resources. Therefore, if we can expect to take notice of students' academic performance and provide guidance to students who cannot pass the threshold as expected, the waste of school resources can effectively be reduced. In this research, the recent years' student data and course results are used as training data to construct student performance prediction models. The K-Means algorithm was used to classify all courses from the freshman to the senior. The related courses will be grouped in the same cluster, which will more likely to find similar features and improve the accuracy of the prediction. Then, this study constructs independent neural networks for each course according to the different academic year. Each model will be pre-trained by using Denoising Auto-encoder. After pre-training, the corresponding structure and weights are taken as the initial value of the neural network model. Each neural network is treated as a base predictor. All predictors will be integrated into an Ensemble predictor according to different years' weights to predict the current student's course performance. As the students finish the course at the end of each semester, the prediction model will continue track and update to enhance model accuracy through online learning.
{"title":"Building Student Course Performance Prediction Model Based on Deep Learning","authors":"J. Kuo, Hao-Ting Chung, Ping-Feng Wang, Baiying Lei","doi":"10.6688/JISE.202101_37(1).0015","DOIUrl":"https://doi.org/10.6688/JISE.202101_37(1).0015","url":null,"abstract":"The deferral of graduation rate in Taiwan's universities is estimated 16%, which will affect the scheduling of school resources. Therefore, if we can expect to take notice of students' academic performance and provide guidance to students who cannot pass the threshold as expected, the waste of school resources can effectively be reduced. In this research, the recent years' student data and course results are used as training data to construct student performance prediction models. The K-Means algorithm was used to classify all courses from the freshman to the senior. The related courses will be grouped in the same cluster, which will more likely to find similar features and improve the accuracy of the prediction. Then, this study constructs independent neural networks for each course according to the different academic year. Each model will be pre-trained by using Denoising Auto-encoder. After pre-training, the corresponding structure and weights are taken as the initial value of the neural network model. Each neural network is treated as a base predictor. All predictors will be integrated into an Ensemble predictor according to different years' weights to predict the current student's course performance. As the students finish the course at the end of each semester, the prediction model will continue track and update to enhance model accuracy through online learning.","PeriodicalId":50177,"journal":{"name":"Journal of Information Science and Engineering","volume":null,"pages":null},"PeriodicalIF":1.1,"publicationDate":"2021-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"73263857","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-01-01DOI: 10.6688/JISE.202101_37(1).0009
Shang-Pin Ma, Chi-Chia Li, Shin-Jie Lee, Hsi-Min Chen, Wen-Tin Lee
SHANG-PIN MA, CHI-CHIA LI, SHIN-JIE LEE, HSI-MIN CHEN, WEN-TIN LEE Department of Computer Science and Engineering National Taiwan Ocean University Keelung 202, Taiwan Department of Information Engineering and Computer Science, National Cheng Kung University, Tainan 701, Taiwan Department of Information Engineering and Computer Science, Feng Chia University, Taiching 407, Taiwan Department of Software Engineering and Management, National Kaohsiung Normal University, Kaohsiung 802, Taiwan E-mail: albert@ntou.edu.tw (Shang-Pin Ma)
{"title":"Cache-Enabled and Context-Aware Approach to Building Composite Mobile Apps","authors":"Shang-Pin Ma, Chi-Chia Li, Shin-Jie Lee, Hsi-Min Chen, Wen-Tin Lee","doi":"10.6688/JISE.202101_37(1).0009","DOIUrl":"https://doi.org/10.6688/JISE.202101_37(1).0009","url":null,"abstract":"SHANG-PIN MA, CHI-CHIA LI, SHIN-JIE LEE, HSI-MIN CHEN, WEN-TIN LEE Department of Computer Science and Engineering National Taiwan Ocean University Keelung 202, Taiwan Department of Information Engineering and Computer Science, National Cheng Kung University, Tainan 701, Taiwan Department of Information Engineering and Computer Science, Feng Chia University, Taiching 407, Taiwan Department of Software Engineering and Management, National Kaohsiung Normal University, Kaohsiung 802, Taiwan E-mail: albert@ntou.edu.tw (Shang-Pin Ma)","PeriodicalId":50177,"journal":{"name":"Journal of Information Science and Engineering","volume":null,"pages":null},"PeriodicalIF":1.1,"publicationDate":"2021-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"73286334","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-01-01DOI: 10.6688/JISE.202109_37(5).0012
Sandro Domitran, Marina Bagić Babac
{"title":"The Use of Deep Reinforcement Learning for Flying a Drone","authors":"Sandro Domitran, Marina Bagić Babac","doi":"10.6688/JISE.202109_37(5).0012","DOIUrl":"https://doi.org/10.6688/JISE.202109_37(5).0012","url":null,"abstract":"","PeriodicalId":50177,"journal":{"name":"Journal of Information Science and Engineering","volume":null,"pages":null},"PeriodicalIF":1.1,"publicationDate":"2021-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"82662426","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-12-02DOI: 10.32674/JISE.V10I1.2396
E. Archer, Yuqian Zhang
“We are on the precipice of an epoch,” in which 21st century organizations are facing a complex, competitive landscape driven largely by globalization and the technological revolution (Hitt, 1998, p. 218). As such, Bikson, Treverton, Moini and Lindstrom (2003) have urged universities to develop a global leadership curriculum, based on their prediction of a future shortage of global leaders in all sectors. This essay examines the critical role of global and culturally responsive leadership for graduates of higher education institutions.
{"title":"Critical Role of Global and Culturally Responsive Leadership for Higher Education in the 21st-Century","authors":"E. Archer, Yuqian Zhang","doi":"10.32674/JISE.V10I1.2396","DOIUrl":"https://doi.org/10.32674/JISE.V10I1.2396","url":null,"abstract":"“We are on the precipice of an epoch,” in which 21st century organizations are facing a complex, competitive landscape driven largely by globalization and the technological revolution (Hitt, 1998, p. 218). As such, Bikson, Treverton, Moini and Lindstrom (2003) have urged universities to develop a global leadership curriculum, based on their prediction of a future shortage of global leaders in all sectors. This essay examines the critical role of global and culturally responsive leadership for graduates of higher education institutions.","PeriodicalId":50177,"journal":{"name":"Journal of Information Science and Engineering","volume":null,"pages":null},"PeriodicalIF":1.1,"publicationDate":"2020-12-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"44571198","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-11-01DOI: 10.6688/JISE.202011_36(6).0015
R. Mehta
Effective data compression and sustainable management of limited bandwidth are the fundamental challenges in the design of current ad-hoc wireless networks. Dynamic Huffman coding protocol with optimal prefix-free codewords assigned to network parameters through cross-layer methodology can considerably address these issues for performance enhancement in these networks. In this paper, the entropy based Huffman binary and ternary coding schemes are implemented with cross-layer design in mobile ad-hoc networks. This cross-layer approach for achieving higher adaptivity incorporated four layers of the traditional networking stack. The simulation results demonstrated the performance tradeoff between throughput and coding metrics for the proposed cross-layer architecture. Moreover, our proposed model exhibits substantially higher compression ratio and coding efficiency than the other existing methods through the integration of this lossless coding protocol in the developed cross-layer framework.
{"title":"Optimal Huffman Coding Performance of Ad-Hoc Networks Based on Cross-Layer Design","authors":"R. Mehta","doi":"10.6688/JISE.202011_36(6).0015","DOIUrl":"https://doi.org/10.6688/JISE.202011_36(6).0015","url":null,"abstract":"Effective data compression and sustainable management of limited bandwidth are the fundamental challenges in the design of current ad-hoc wireless networks. Dynamic Huffman coding protocol with optimal prefix-free codewords assigned to network parameters through cross-layer methodology can considerably address these issues for performance enhancement in these networks. In this paper, the entropy based Huffman binary and ternary coding schemes are implemented with cross-layer design in mobile ad-hoc networks. This cross-layer approach for achieving higher adaptivity incorporated four layers of the traditional networking stack. The simulation results demonstrated the performance tradeoff between throughput and coding metrics for the proposed cross-layer architecture. Moreover, our proposed model exhibits substantially higher compression ratio and coding efficiency than the other existing methods through the integration of this lossless coding protocol in the developed cross-layer framework.","PeriodicalId":50177,"journal":{"name":"Journal of Information Science and Engineering","volume":null,"pages":null},"PeriodicalIF":1.1,"publicationDate":"2020-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"89708757","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-11-01DOI: 10.6688/JISE.202011_36(6).0008
Ali Algarni, A. Almarashi, G. Abd-Elmougod
Products come from different lines with the same facility are tested under comparative life tests which known with the jointly censoring scheme. In this paper, two sets of products under the same facility have Weibull lifetime distributions are selected to test under Type- I generalized hybrid censoring scheme (GHCS). The observed censoring data are used to build the maximum likelihood (ML) estimators as well as approximate confidence intervals for the model parameters. Also, Bayes estimators with the help of MCMC methods are discussed. The analysis of simulated data set with Monte Carlo simulation study is used to illustrate and compare the theoretical results. Finally, a brief comment is summarized in concluding section.
{"title":"Joint Type-I Generalized Hybrid Censoring for Estimation Two Weibull Distributions","authors":"Ali Algarni, A. Almarashi, G. Abd-Elmougod","doi":"10.6688/JISE.202011_36(6).0008","DOIUrl":"https://doi.org/10.6688/JISE.202011_36(6).0008","url":null,"abstract":"Products come from different lines with the same facility are tested under comparative life tests which known with the jointly censoring scheme. In this paper, two sets of products under the same facility have Weibull lifetime distributions are selected to test under Type- I generalized hybrid censoring scheme (GHCS). The observed censoring data are used to build the maximum likelihood (ML) estimators as well as approximate confidence intervals for the model parameters. Also, Bayes estimators with the help of MCMC methods are discussed. The analysis of simulated data set with Monte Carlo simulation study is used to illustrate and compare the theoretical results. Finally, a brief comment is summarized in concluding section.","PeriodicalId":50177,"journal":{"name":"Journal of Information Science and Engineering","volume":null,"pages":null},"PeriodicalIF":1.1,"publicationDate":"2020-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"83948181","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-11-01DOI: 10.6688/JISE.202011_36(6).0011
Tayyab Ahmed Shaikh, Syed Sajjad Hussain, M. Tanweer, M. Hashmani
In this paper, a new algorithm incorporating broadening selection strategy in competitive constraint handling paradigm for finding the optimum solution in constrained problems has been proposed, referred as Broadening Selection Competitive Constraint Handling (BSCCH). Although, competitive constraint handling approaches have proved to be very efficient, but they lack faster convergence due to offspring generation from random individuals. By incorporating selection strategy such as broadening selection in the competitive approach, better results are obtained and convergence rate is improved significantly. Incorporating said strategy, the BSCCH algorithm has been proposed which is generic in nature and can be coupled with various evolutionary algorithms. In this study, the BSCCH algorithm has been coupled with Differential Evolution algorithm as a proof of concept because it is found to be an efficient algorithm in the literature for constrained optimization problems. The proposed algorithm has been evaluated using 24 benchmark functions. The mean closure performance of the BSCCH algorithm is compared against seven selected state-of-the-art algorithms, namely Differential Evolution with Adaptive Trial Vector Generation Strategy and Cluster-replacement-based Feasibility Rule (CACDE), Improved Teaching Learning Based Optimization (ITLBO), Modified Global Best Artificial Bee Colony (MGABC), Stochastic Ranking Differential Evolution (SRDE), Novel Differential Evolution (NDE), Partical Swarm Optimization for solving engineering problems - a new constraint handling mechanism (CVI-PSO) and Ensemble of Constraint Handling Techniques (ECHT). The median convergence traces have been compared with two different algorithms based on differential evolution, i:e: Ensemble of Constraint Handling Techniques (ECHT) and Stochastic Ranking Differential Evolution (SRDE). ECHT is considered to be a flagship ensemble technique till date for constrained optimization problems, whereas SRDE employs a parent selection mechanism for constrained optimization. The proposed algorithm is found to provide better solutions and achieve significantly faster convergence in most of the problems.
{"title":"Broadening Selection Competitive Constraint Handling Algorithm for Faster Convergence","authors":"Tayyab Ahmed Shaikh, Syed Sajjad Hussain, M. Tanweer, M. Hashmani","doi":"10.6688/JISE.202011_36(6).0011","DOIUrl":"https://doi.org/10.6688/JISE.202011_36(6).0011","url":null,"abstract":"In this paper, a new algorithm incorporating broadening selection strategy in competitive constraint handling paradigm for finding the optimum solution in constrained problems has been proposed, referred as Broadening Selection Competitive Constraint Handling (BSCCH). Although, competitive constraint handling approaches have proved to be very efficient, but they lack faster convergence due to offspring generation from random individuals. By incorporating selection strategy such as broadening selection in the competitive approach, better results are obtained and convergence rate is improved significantly. Incorporating said strategy, the BSCCH algorithm has been proposed which is generic in nature and can be coupled with various evolutionary algorithms. In this study, the BSCCH algorithm has been coupled with Differential Evolution algorithm as a proof of concept because it is found to be an efficient algorithm in the literature for constrained optimization problems. The proposed algorithm has been evaluated using 24 benchmark functions. The mean closure performance of the BSCCH algorithm is compared against seven selected state-of-the-art algorithms, namely Differential Evolution with Adaptive Trial Vector Generation Strategy and Cluster-replacement-based Feasibility Rule (CACDE), Improved Teaching Learning Based Optimization (ITLBO), Modified Global Best Artificial Bee Colony (MGABC), Stochastic Ranking Differential Evolution (SRDE), Novel Differential Evolution (NDE), Partical Swarm Optimization for solving engineering problems - a new constraint handling mechanism (CVI-PSO) and Ensemble of Constraint Handling Techniques (ECHT). The median convergence traces have been compared with two different algorithms based on differential evolution, i:e: Ensemble of Constraint Handling Techniques (ECHT) and Stochastic Ranking Differential Evolution (SRDE). ECHT is considered to be a flagship ensemble technique till date for constrained optimization problems, whereas SRDE employs a parent selection mechanism for constrained optimization. The proposed algorithm is found to provide better solutions and achieve significantly faster convergence in most of the problems.","PeriodicalId":50177,"journal":{"name":"Journal of Information Science and Engineering","volume":null,"pages":null},"PeriodicalIF":1.1,"publicationDate":"2020-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"72854919","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-11-01DOI: 10.6688/JISE.202011_36(6).0009
N. AlShamrani, A. Elaiw, H. Batarfi, A. Hobiny
In this work we present two mathematical models for the infection dynamics of scabies. The dynamics is described by four-dimensional system of ordinary differential equations that expresses the transmissions between susceptible and infectious/infective individuals. In the second model, we include the importance of adult scabiei mite in the real interaction with hosts. Nonnegativity and boundedness of solutions of the models are conducted. A threshold parameter is calculated for each model which ensures the existence of all corresponding equilibria. Using candidate Lyapunov functions, it is shown that whenever the threshold parameter is less than or equal unity, the models have an associated disease-free equilibrium that is globally asymptotically stable. In addition, when the threshold exceeds unity the models have a globally asymptotically stable endemic equilibrium. Finally, using some parameter values related to the scabies infection dynamics, numerical simulation results are demonstrated to clarify the main theoretical results.
{"title":"Modeling and Analysis of Scabies Transmission Disease","authors":"N. AlShamrani, A. Elaiw, H. Batarfi, A. Hobiny","doi":"10.6688/JISE.202011_36(6).0009","DOIUrl":"https://doi.org/10.6688/JISE.202011_36(6).0009","url":null,"abstract":"In this work we present two mathematical models for the infection dynamics of scabies. The dynamics is described by four-dimensional system of ordinary differential equations that expresses the transmissions between susceptible and infectious/infective individuals. In the second model, we include the importance of adult scabiei mite in the real interaction with hosts. Nonnegativity and boundedness of solutions of the models are conducted. A threshold parameter is calculated for each model which ensures the existence of all corresponding equilibria. Using candidate Lyapunov functions, it is shown that whenever the threshold parameter is less than or equal unity, the models have an associated disease-free equilibrium that is globally asymptotically stable. In addition, when the threshold exceeds unity the models have a globally asymptotically stable endemic equilibrium. Finally, using some parameter values related to the scabies infection dynamics, numerical simulation results are demonstrated to clarify the main theoretical results.","PeriodicalId":50177,"journal":{"name":"Journal of Information Science and Engineering","volume":null,"pages":null},"PeriodicalIF":1.1,"publicationDate":"2020-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"82010228","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-11-01DOI: 10.6688/JISE.202011_36(6).0010
A. Abdel‐Aty, M. Khater, A. Zidan, R. Attia
This research employs a new analytical scheme to construct novel traveling wave solutions of the Wick-type stochastic Schamel KdV equation. This equation explains the electrostatic potential for a particular electron distribution in velocity space. It is also used to explain the nonlinear interaction of ion-acoustic waves when electron trapping. By using the Hermite transform, inverse Hermite transforms, and white noise analysis allows us for applying the modified Khater method to this model. Many novel solutions are obtained and sketched to discuss more physical properties of the model.
{"title":"New Analytical Solutions of Wick-Type Stochastic Schamel KdV Equation Via Modified Khater Method","authors":"A. Abdel‐Aty, M. Khater, A. Zidan, R. Attia","doi":"10.6688/JISE.202011_36(6).0010","DOIUrl":"https://doi.org/10.6688/JISE.202011_36(6).0010","url":null,"abstract":"This research employs a new analytical scheme to construct novel traveling wave solutions of the Wick-type stochastic Schamel KdV equation. This equation explains the electrostatic potential for a particular electron distribution in velocity space. It is also used to explain the nonlinear interaction of ion-acoustic waves when electron trapping. By using the Hermite transform, inverse Hermite transforms, and white noise analysis allows us for applying the modified Khater method to this model. Many novel solutions are obtained and sketched to discuss more physical properties of the model.","PeriodicalId":50177,"journal":{"name":"Journal of Information Science and Engineering","volume":null,"pages":null},"PeriodicalIF":1.1,"publicationDate":"2020-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"78988652","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-11-01DOI: 10.6688/JISE.202011_36(6).0013
Hong-Ji Wang, Xiang Xu, Baomin Xu, Yu Shuang-Yuan, Wang Quan-Xin
In the context of model compression using the student-teacher paradigm, we propose the idea of student-centric learning, where the student is less constrained by the teacher and able to learn on its own. We believe the student should have more flexibility during training. Towards student-centric learning, we propose two approaches: correlation-based learning and self-guided learning. In correlation-based learning, we propose to guide the student with two types of correlations between activations: the correlation between different channels and the correlation between different spatial locations. In self-guided learning, we propose to give the student network the opportunity to learn by itself in the form of additional self-taught neurons. We empirically validate our approaches on benchmark datasets, producing state-of-the-art results. Notably, our approaches can train a smaller and shallower student network with only 5 layers that outperforms a larger and deeper teacher network with 11 layers by nearly 1% on CIFAR-100.
{"title":"Student-Centric Network Learning for Improved Knowledge Transfer","authors":"Hong-Ji Wang, Xiang Xu, Baomin Xu, Yu Shuang-Yuan, Wang Quan-Xin","doi":"10.6688/JISE.202011_36(6).0013","DOIUrl":"https://doi.org/10.6688/JISE.202011_36(6).0013","url":null,"abstract":"In the context of model compression using the student-teacher paradigm, we propose the idea of student-centric learning, where the student is less constrained by the teacher and able to learn on its own. We believe the student should have more flexibility during training. Towards student-centric learning, we propose two approaches: correlation-based learning and self-guided learning. In correlation-based learning, we propose to guide the student with two types of correlations between activations: the correlation between different channels and the correlation between different spatial locations. In self-guided learning, we propose to give the student network the opportunity to learn by itself in the form of additional self-taught neurons. We empirically validate our approaches on benchmark datasets, producing state-of-the-art results. Notably, our approaches can train a smaller and shallower student network with only 5 layers that outperforms a larger and deeper teacher network with 11 layers by nearly 1% on CIFAR-100.","PeriodicalId":50177,"journal":{"name":"Journal of Information Science and Engineering","volume":null,"pages":null},"PeriodicalIF":1.1,"publicationDate":"2020-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"82010397","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：