Pub Date : 2020-03-01DOI: 10.1109/ICICT50521.2020.00017
Hiroki Watanabe, T. Malý, Johannes Wallner, G. Prokop
In terms of vehicle safety, the number of advanced driver assistance systems (ADAS) mounted in an automobile has been increasing recently. For an efficient conceptional design and system validation of ADAS, the representative test scenarios are indispensable. In order to identify the representative scenarios,the real-world traffic scenarios are to be clustered according to their similarity. The hierarchical agglomerative clustering is a well-known method to quantify the similarity of traffic scenarios existing in a database. However, the cluster structure is affected by the linkage criterion used in the agglomerative procedure.This study inquires into the similarity measurement of vehicle-pedestrian near-crashes in the USA. Various linkage criteria are selected to get better understanding of their influence on the clustering results and conduct a comparative study. Furthermore,a hybrid clustering algorithm is presented, which is based on k-covers and k-means clustering. Using the average silhouette width, the optimal number of clusters is calculated and the cluster structures are investigated. In the end, the representative scenarios are selected with the use of centrality measure and form the basis of the scenario catalog making for the reduction of test effort in ADAS development.
{"title":"Cluster-Linkage Analysis in Traffic Data Clustering for Development of Advanced Driver Assistance Systems","authors":"Hiroki Watanabe, T. Malý, Johannes Wallner, G. Prokop","doi":"10.1109/ICICT50521.2020.00017","DOIUrl":"https://doi.org/10.1109/ICICT50521.2020.00017","url":null,"abstract":"In terms of vehicle safety, the number of advanced driver assistance systems (ADAS) mounted in an automobile has been increasing recently. For an efficient conceptional design and system validation of ADAS, the representative test scenarios are indispensable. In order to identify the representative scenarios,the real-world traffic scenarios are to be clustered according to their similarity. The hierarchical agglomerative clustering is a well-known method to quantify the similarity of traffic scenarios existing in a database. However, the cluster structure is affected by the linkage criterion used in the agglomerative procedure.This study inquires into the similarity measurement of vehicle-pedestrian near-crashes in the USA. Various linkage criteria are selected to get better understanding of their influence on the clustering results and conduct a comparative study. Furthermore,a hybrid clustering algorithm is presented, which is based on k-covers and k-means clustering. Using the average silhouette width, the optimal number of clusters is calculated and the cluster structures are investigated. In the end, the representative scenarios are selected with the use of centrality measure and form the basis of the scenario catalog making for the reduction of test effort in ADAS development.","PeriodicalId":445000,"journal":{"name":"2020 3rd International Conference on Information and Computer Technologies (ICICT)","volume":"43 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134338224","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-03-01DOI: 10.1109/ICICT50521.2020.00092
J. Yeh, Craig W. Bartholio, Elyse Shackleton, Levi Costello, Matthew Perera, Kyle Yeh, Chelsea Yeh
In this paper, we describe the application of the technologies of the Internet of Things (IoT) to secondary and higher education. Specifically, it seeks to monitor the status of the students and estimate student attention and engagement based on data collected by environmentally embedded IoT sensors, and provide real-time feedback to the instructor. The research was divided into three parts. The first part researched the IoT sensors and devices used to capture student physiological data. The second part developed the instructor feedback devices, which included tablet and smartwatch applications. The third part is the creation of the machine learning model that performed the estimation of student attention from the data.
{"title":"Environmentally Embedded Internet-of-Things for Secondary and Higher Education","authors":"J. Yeh, Craig W. Bartholio, Elyse Shackleton, Levi Costello, Matthew Perera, Kyle Yeh, Chelsea Yeh","doi":"10.1109/ICICT50521.2020.00092","DOIUrl":"https://doi.org/10.1109/ICICT50521.2020.00092","url":null,"abstract":"In this paper, we describe the application of the technologies of the Internet of Things (IoT) to secondary and higher education. Specifically, it seeks to monitor the status of the students and estimate student attention and engagement based on data collected by environmentally embedded IoT sensors, and provide real-time feedback to the instructor. The research was divided into three parts. The first part researched the IoT sensors and devices used to capture student physiological data. The second part developed the instructor feedback devices, which included tablet and smartwatch applications. The third part is the creation of the machine learning model that performed the estimation of student attention from the data.","PeriodicalId":445000,"journal":{"name":"2020 3rd International Conference on Information and Computer Technologies (ICICT)","volume":"9 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130975279","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-03-01DOI: 10.1109/ICICT50521.2020.00028
Nawar H. Obeidat, C. Purdy
Designing smart systems has recently received much attention because of the ability to connect heterogeneous sub-systems together to build a fully controlled smart system. In this project, designing a smart, safe, secure school system for students is our eventual goal. In this paper, we propose a model for a smart school building system. We model the system using UML diagrams to illustrate the system's objects and sequences of actions. We use formal methods techniques to verify the properties of the system. Specifically, we use TLA+, the formal specification language, for our formal modeling. As for the smart school building, we design the school to be safe by requiring a username and password for entering the building. The system can also control the lighting system and the temperature for the whole building. For additional safety, the system has a smoke detection sub-system in case of fire. We propose this smart school building system and verify its properties and correctness using the TLA+ toolbox.
{"title":"Modeling a Smart School Building System Using UML and TLA+","authors":"Nawar H. Obeidat, C. Purdy","doi":"10.1109/ICICT50521.2020.00028","DOIUrl":"https://doi.org/10.1109/ICICT50521.2020.00028","url":null,"abstract":"Designing smart systems has recently received much attention because of the ability to connect heterogeneous sub-systems together to build a fully controlled smart system. In this project, designing a smart, safe, secure school system for students is our eventual goal. In this paper, we propose a model for a smart school building system. We model the system using UML diagrams to illustrate the system's objects and sequences of actions. We use formal methods techniques to verify the properties of the system. Specifically, we use TLA+, the formal specification language, for our formal modeling. As for the smart school building, we design the school to be safe by requiring a username and password for entering the building. The system can also control the lighting system and the temperature for the whole building. For additional safety, the system has a smoke detection sub-system in case of fire. We propose this smart school building system and verify its properties and correctness using the TLA+ toolbox.","PeriodicalId":445000,"journal":{"name":"2020 3rd International Conference on Information and Computer Technologies (ICICT)","volume":"10 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133387742","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-03-01DOI: 10.1109/ICICT50521.2020.00021
Pranjal Ranjan, Sarvesh Patil, F. Kazi
Deep-Fakes are emerging as a significant threat to society, with potential to become weapons of mass disinformation and chaos. Simple tools provide ways to produce such digital forgeries at a large scale which makes it crucial to develop counter-attacking approaches for detection of these Deep-Learning based manipulations. This work analyzes a Transfer Learning based Convolutional Neural Network framework for the task of Deep-Fake Detection on three of the latest released datasets – DeepFakeDetection (DFD), Celeb-DF, and DeepFakeDetectionChallenge (DFDC) Preview. Additionally, a custom dataset of high-quality Deep-Fakes is compiled and used for evaluation of models. The intuition behind Transfer Learning for Deep-Fakes Detection is explored using the Explainable-AI technique of visualizing intermediate activations to provide interpretability. The critical problem of dataset shift and its effect on domain adaptation is explored by comparing cross-dataset test accuracies, with and without the usage of Transfer Learning. The results of this work indicate that even though Deep-Fake Detection is a highly domain specific task, there is a significant improvement in performance in terms of both single-domain classification accuracy and generalizability by utilizing Transfer Learning.
{"title":"Improved Generalizability of Deep-Fakes Detection using Transfer Learning Based CNN Framework","authors":"Pranjal Ranjan, Sarvesh Patil, F. Kazi","doi":"10.1109/ICICT50521.2020.00021","DOIUrl":"https://doi.org/10.1109/ICICT50521.2020.00021","url":null,"abstract":"Deep-Fakes are emerging as a significant threat to society, with potential to become weapons of mass disinformation and chaos. Simple tools provide ways to produce such digital forgeries at a large scale which makes it crucial to develop counter-attacking approaches for detection of these Deep-Learning based manipulations. This work analyzes a Transfer Learning based Convolutional Neural Network framework for the task of Deep-Fake Detection on three of the latest released datasets – DeepFakeDetection (DFD), Celeb-DF, and DeepFakeDetectionChallenge (DFDC) Preview. Additionally, a custom dataset of high-quality Deep-Fakes is compiled and used for evaluation of models. The intuition behind Transfer Learning for Deep-Fakes Detection is explored using the Explainable-AI technique of visualizing intermediate activations to provide interpretability. The critical problem of dataset shift and its effect on domain adaptation is explored by comparing cross-dataset test accuracies, with and without the usage of Transfer Learning. The results of this work indicate that even though Deep-Fake Detection is a highly domain specific task, there is a significant improvement in performance in terms of both single-domain classification accuracy and generalizability by utilizing Transfer Learning.","PeriodicalId":445000,"journal":{"name":"2020 3rd International Conference on Information and Computer Technologies (ICICT)","volume":"53 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114563288","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-03-01DOI: 10.1109/ICICT50521.2020.00071
Daechul Park, Youngeun Cho, Chang-Gun Lee
Throughout the last decade, the importance of parallel computing has risen greatly to match the ever-increasing computational demand. Frameworks such as OpenMP and OpenCL allow easy parallelization of computing tasks into desirable number of threads, opening up a chance to greatly utilize the parallel computing resources. We call this "parallelization freedom". However, this does not come for free, as parallelization overhead increase with parallelization option (i.e. the number of thread each task is parallelized). Thus parallelization option must be carefully decided to better utilize a given computing resource. This paper addresses the problem of assigning parallelization option to each task for global FP scheduler. For this, we extend the approaches made by Cho, which is limited to the global EDF scheduler case. We prove that a conditionally optimal parallelization assignment of parallelization option also exists for the global FP case. Through extensive simulations, we show a significant improvement of schedulability.
{"title":"Conditionally Optimal Parallelization for Global FP on Multi-core Systems","authors":"Daechul Park, Youngeun Cho, Chang-Gun Lee","doi":"10.1109/ICICT50521.2020.00071","DOIUrl":"https://doi.org/10.1109/ICICT50521.2020.00071","url":null,"abstract":"Throughout the last decade, the importance of parallel computing has risen greatly to match the ever-increasing computational demand. Frameworks such as OpenMP and OpenCL allow easy parallelization of computing tasks into desirable number of threads, opening up a chance to greatly utilize the parallel computing resources. We call this \"parallelization freedom\". However, this does not come for free, as parallelization overhead increase with parallelization option (i.e. the number of thread each task is parallelized). Thus parallelization option must be carefully decided to better utilize a given computing resource. This paper addresses the problem of assigning parallelization option to each task for global FP scheduler. For this, we extend the approaches made by Cho, which is limited to the global EDF scheduler case. We prove that a conditionally optimal parallelization assignment of parallelization option also exists for the global FP case. Through extensive simulations, we show a significant improvement of schedulability.","PeriodicalId":445000,"journal":{"name":"2020 3rd International Conference on Information and Computer Technologies (ICICT)","volume":"33 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124896706","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-03-01DOI: 10.1109/ICICT50521.2020.00014
Giovanni Ortegón Cortázar, O. Garcia-Bedoya, J. Samper, Francisco García-Sánchez
One of the objectives of the fourth industrial revolution is the use of data to make decisions, in this sense the decisions of the government could be better if they are based on useful data sources, with quality and focused on services to citizens. This article presents a case of use in the government of Colombia in which an information system was created for the management of complaints and claims by means of control panels, visualization of statistical information and high personalization in the processes for more than 1200 offices and 8000 employees.
{"title":"A Case of Use in the Management of Administrative Processes, Control of Procedures and Services to Citizens In E-Government","authors":"Giovanni Ortegón Cortázar, O. Garcia-Bedoya, J. Samper, Francisco García-Sánchez","doi":"10.1109/ICICT50521.2020.00014","DOIUrl":"https://doi.org/10.1109/ICICT50521.2020.00014","url":null,"abstract":"One of the objectives of the fourth industrial revolution is the use of data to make decisions, in this sense the decisions of the government could be better if they are based on useful data sources, with quality and focused on services to citizens. This article presents a case of use in the government of Colombia in which an information system was created for the management of complaints and claims by means of control panels, visualization of statistical information and high personalization in the processes for more than 1200 offices and 8000 employees.","PeriodicalId":445000,"journal":{"name":"2020 3rd International Conference on Information and Computer Technologies (ICICT)","volume":"87 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127180148","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-01-27DOI: 10.1109/ICICT50521.2020.00065
Hardik B. Jain, M. Edwards, Ethan R. Elenberg, A. Rawat, S. Vishwanath
Many performance critical systems today must rely on performance enhancements, such as multi-port memories, to keep up with the increasing demand of memory-access capacity. However, the large area footprints and complexity of existing multi-port memory designs limit their applicability. This paper explores a coding theoretic framework to address this problem. In particular, this paper introduces a framework to encode data across multiple single-port memory banks in order to algorithmically realize the functionality of multi-port memory. This paper proposes three code designs with significantly less storage overhead compared to the existing replication based emulations of multi-port memories. To further improve performance, we also demonstrate a memory controller design that utilizes redundancy across coded memory banks to more efficiently schedule read and write requests sent across multiple cores. Furthermore, guided by DRAM traces, the paper explores dynamic coding techniques to improve the efficiency of the coding based memory design. We then show significant performance improvements in critical word read and write latency in the proposed coded-memory design when compared to a traditional uncoded-memory design.
{"title":"Achieving Multi-port Memory Performance on Single-Port Memory with Coding Techniques","authors":"Hardik B. Jain, M. Edwards, Ethan R. Elenberg, A. Rawat, S. Vishwanath","doi":"10.1109/ICICT50521.2020.00065","DOIUrl":"https://doi.org/10.1109/ICICT50521.2020.00065","url":null,"abstract":"Many performance critical systems today must rely on performance enhancements, such as multi-port memories, to keep up with the increasing demand of memory-access capacity. However, the large area footprints and complexity of existing multi-port memory designs limit their applicability. This paper explores a coding theoretic framework to address this problem. In particular, this paper introduces a framework to encode data across multiple single-port memory banks in order to algorithmically realize the functionality of multi-port memory. This paper proposes three code designs with significantly less storage overhead compared to the existing replication based emulations of multi-port memories. To further improve performance, we also demonstrate a memory controller design that utilizes redundancy across coded memory banks to more efficiently schedule read and write requests sent across multiple cores. Furthermore, guided by DRAM traces, the paper explores dynamic coding techniques to improve the efficiency of the coding based memory design. We then show significant performance improvements in critical word read and write latency in the proposed coded-memory design when compared to a traditional uncoded-memory design.","PeriodicalId":445000,"journal":{"name":"2020 3rd International Conference on Information and Computer Technologies (ICICT)","volume":"47 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-01-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123110720","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-01-13DOI: 10.1109/ICICT50521.2020.00022
M. H. Tanveer, Antony Thomas, Xiaowei Wu, Hongxiao Zhu
In this article, we propose a new approach for simu-lating trees, including their branches, sub-branches, and leaves. This approach combines the theory of biological development, mathematical models, and computer graphics, producing simu-lated trees and forest with full geometry. Specifically, we adopt the Lindenmayer process to simulate the branching pattern of trees and modify the available measurements and dimensions of 3D CAD developed object files to create natural looking sub-branches and leaves. Randomization has been added to the placement of all branches, sub branches and leaves. To simulate a forest, we adopt Inhomogeneous Poisson process to generate random locations of trees. Our approach can be used to create complex structured 3D virtual environment for the purpose of testing new sensors and training robotic algorithms. We look forward to applying this approach to test biosonar sensors that mimick bats' fly in the simulated environment.
{"title":"Simulate Forest Trees by Integrating L-System and 3D CAD Files","authors":"M. H. Tanveer, Antony Thomas, Xiaowei Wu, Hongxiao Zhu","doi":"10.1109/ICICT50521.2020.00022","DOIUrl":"https://doi.org/10.1109/ICICT50521.2020.00022","url":null,"abstract":"In this article, we propose a new approach for simu-lating trees, including their branches, sub-branches, and leaves. This approach combines the theory of biological development, mathematical models, and computer graphics, producing simu-lated trees and forest with full geometry. Specifically, we adopt the Lindenmayer process to simulate the branching pattern of trees and modify the available measurements and dimensions of 3D CAD developed object files to create natural looking sub-branches and leaves. Randomization has been added to the placement of all branches, sub branches and leaves. To simulate a forest, we adopt Inhomogeneous Poisson process to generate random locations of trees. Our approach can be used to create complex structured 3D virtual environment for the purpose of testing new sensors and training robotic algorithms. We look forward to applying this approach to test biosonar sensors that mimick bats' fly in the simulated environment.","PeriodicalId":445000,"journal":{"name":"2020 3rd International Conference on Information and Computer Technologies (ICICT)","volume":"84 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-01-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121235125","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-01-07DOI: 10.1109/ICICT50521.2020.00068
Said Varlioglu, Bilal Gonen, Murat Ozer, Mehmet F. Bastug
Cryptojacking is the exploitation of victims' computer resources to mine for cryptocurrency using malicious scripts. It had become popular after 2017 when attackers started to exploit legal mining scripts, especially Coinhive scripts. Coinhive was actually a legal mining service that provided scripts and servers for in-browser mining activities. Nevertheless, over 10 million web users had been victims every month before the Coinhive shutdown that happened in Mar 2019. This paper explores the new era of the cryptojacking world after Coinhive discontinued its service. We aimed to see whether and how attackers continue cryptojacking, generate new malicious scripts, and developed new methods. We used a capable cryptojacking detector named CMTracker that proposed by Hong et al. in 2018. We automatically and manually examined 2770 websites that had been detected by CMTracker before the Coinhive shutdown. The results revealed that 99% of sites no longer continue cryptojacking. 1% of websites still run 8 unique mining scripts. By tracking these mining scripts, we detected 632 unique cryptojacking websites. Moreover, open-source investigations (OSINT) demonstrated that attackers still use the same methods. Therefore, we listed the typical patterns of cryptojacking. We concluded that cryptojacking is not dead after the Coinhive shutdown. It is still alive, but not as attractive as it used to be.
{"title":"Is Cryptojacking Dead After Coinhive Shutdown?","authors":"Said Varlioglu, Bilal Gonen, Murat Ozer, Mehmet F. Bastug","doi":"10.1109/ICICT50521.2020.00068","DOIUrl":"https://doi.org/10.1109/ICICT50521.2020.00068","url":null,"abstract":"Cryptojacking is the exploitation of victims' computer resources to mine for cryptocurrency using malicious scripts. It had become popular after 2017 when attackers started to exploit legal mining scripts, especially Coinhive scripts. Coinhive was actually a legal mining service that provided scripts and servers for in-browser mining activities. Nevertheless, over 10 million web users had been victims every month before the Coinhive shutdown that happened in Mar 2019. This paper explores the new era of the cryptojacking world after Coinhive discontinued its service. We aimed to see whether and how attackers continue cryptojacking, generate new malicious scripts, and developed new methods. We used a capable cryptojacking detector named CMTracker that proposed by Hong et al. in 2018. We automatically and manually examined 2770 websites that had been detected by CMTracker before the Coinhive shutdown. The results revealed that 99% of sites no longer continue cryptojacking. 1% of websites still run 8 unique mining scripts. By tracking these mining scripts, we detected 632 unique cryptojacking websites. Moreover, open-source investigations (OSINT) demonstrated that attackers still use the same methods. Therefore, we listed the typical patterns of cryptojacking. We concluded that cryptojacking is not dead after the Coinhive shutdown. It is still alive, but not as attractive as it used to be.","PeriodicalId":445000,"journal":{"name":"2020 3rd International Conference on Information and Computer Technologies (ICICT)","volume":"25 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-01-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133935861","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-09-12DOI: 10.1109/ICICT50521.2020.00020
Masanari Kimura, Masayuki Tanaka
Deep neural networks (DNNs) are known as black-box models. In other words, it is difficult to interpret the internal state of the model. Improving the interpretability of DNNs is one of the hot research topics. However, at present, the definition of interpretability for DNNs is vague, and the question of what is a highly explanatory model is still controversial. To address this issue, we provide the definition of the human predictability of the model, as a part of the interpretability of the DNNs. The human predictability proposed in this paper is defined by easiness to predict the change of the inference when perturbating the model of the DNNs. In addition, we introduce one example of high human-predictable DNNs. We discuss that our definition will help to the research of the interpretability of the DNNs considering various types of applications.
{"title":"New Perspective of Interpretability of Deep Neural Networks","authors":"Masanari Kimura, Masayuki Tanaka","doi":"10.1109/ICICT50521.2020.00020","DOIUrl":"https://doi.org/10.1109/ICICT50521.2020.00020","url":null,"abstract":"Deep neural networks (DNNs) are known as black-box models. In other words, it is difficult to interpret the internal state of the model. Improving the interpretability of DNNs is one of the hot research topics. However, at present, the definition of interpretability for DNNs is vague, and the question of what is a highly explanatory model is still controversial. To address this issue, we provide the definition of the human predictability of the model, as a part of the interpretability of the DNNs. The human predictability proposed in this paper is defined by easiness to predict the change of the inference when perturbating the model of the DNNs. In addition, we introduce one example of high human-predictable DNNs. We discuss that our definition will help to the research of the interpretability of the DNNs considering various types of applications.","PeriodicalId":445000,"journal":{"name":"2020 3rd International Conference on Information and Computer Technologies (ICICT)","volume":"19 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-09-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132820532","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","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：