Pub Date : 2020-12-01DOI: 10.1109/UCC48980.2020.00063
Jie Zhang, Yi Zhong, Yi Han, Dongdong Li, Chenxi Yu, Junchang Mo
As an emerging multimedia service, virtual reality (VR) video streaming is facing two challenges: extremely large bandwidth requirements and strict delay requirements. There-fore, improving the utilization of network resources is of great significance to the application and development of VR video in order to provide a better quality of experience. Currently, many VR video streaming solutions are based on 360° video, which is a compromise between restricted bandwidth and streaming delay requirement. For a tile-based 360° video streaming using HTTP2 protocol, the video is split into different tiles with multiple quality levels. In the case of insufficient bandwidth, a limited number of quality levels will lead to large quality differences between adjacent zones, which will also limit the optimization of quality level adaptation, resulting in a lower QoE. In this paper, we propose a new tile-based 360° video adaptive streaming scheme based on multiple video quality levels. The proposed method provides more video quality options/levels, dividing the 360° video into different video zones according to its viewpoint position and assign them with different video quality levels based on bandwidth conditions during the streaming time, to achieve smooth video quality distribution, and ensure high QoE with high video bitrate, low-quality switches and minimized stall time. The experimental results show that the QoE of this proposed method is improved by approximately 28% compared with the existing adaptive 360° video streaming scheme.
{"title":"A 360° Video Adaptive Streaming Scheme Based on Multiple Video Qualities","authors":"Jie Zhang, Yi Zhong, Yi Han, Dongdong Li, Chenxi Yu, Junchang Mo","doi":"10.1109/UCC48980.2020.00063","DOIUrl":"https://doi.org/10.1109/UCC48980.2020.00063","url":null,"abstract":"As an emerging multimedia service, virtual reality (VR) video streaming is facing two challenges: extremely large bandwidth requirements and strict delay requirements. There-fore, improving the utilization of network resources is of great significance to the application and development of VR video in order to provide a better quality of experience. Currently, many VR video streaming solutions are based on 360° video, which is a compromise between restricted bandwidth and streaming delay requirement. For a tile-based 360° video streaming using HTTP2 protocol, the video is split into different tiles with multiple quality levels. In the case of insufficient bandwidth, a limited number of quality levels will lead to large quality differences between adjacent zones, which will also limit the optimization of quality level adaptation, resulting in a lower QoE. In this paper, we propose a new tile-based 360° video adaptive streaming scheme based on multiple video quality levels. The proposed method provides more video quality options/levels, dividing the 360° video into different video zones according to its viewpoint position and assign them with different video quality levels based on bandwidth conditions during the streaming time, to achieve smooth video quality distribution, and ensure high QoE with high video bitrate, low-quality switches and minimized stall time. The experimental results show that the QoE of this proposed method is improved by approximately 28% compared with the existing adaptive 360° video streaming scheme.","PeriodicalId":125849,"journal":{"name":"2020 IEEE/ACM 13th International Conference on Utility and Cloud Computing (UCC)","volume":"12 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125222458","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 : 2020-12-01DOI: 10.1109/UCC48980.2020.00070
R. Saleem, Bo Yuan, Fatih Kurugollu, A. Anjum
Artificial Intelligence (AI) models can learn from data and make decisions without any human intervention. However, the deployment of such models is challenging and risky because we do not know how the internal decisionmaking is happening in these models. Especially, the high-risk decisions such as medical diagnosis or automated navigation demand explainability and verification of the decision making process in AI algorithms. This research paper aims to explain Artificial Intelligence (AI) models by discretizing the black-box process model of deep neural networks using partial differential equations. The PDEs based deterministic models would minimize the time and computational cost of the decision-making process and reduce the chances of uncertainty that make the prediction more trustworthy.
{"title":"Explaining probabilistic Artificial Intelligence (AI) models by discretizing Deep Neural Networks","authors":"R. Saleem, Bo Yuan, Fatih Kurugollu, A. Anjum","doi":"10.1109/UCC48980.2020.00070","DOIUrl":"https://doi.org/10.1109/UCC48980.2020.00070","url":null,"abstract":"Artificial Intelligence (AI) models can learn from data and make decisions without any human intervention. However, the deployment of such models is challenging and risky because we do not know how the internal decisionmaking is happening in these models. Especially, the high-risk decisions such as medical diagnosis or automated navigation demand explainability and verification of the decision making process in AI algorithms. This research paper aims to explain Artificial Intelligence (AI) models by discretizing the black-box process model of deep neural networks using partial differential equations. The PDEs based deterministic models would minimize the time and computational cost of the decision-making process and reduce the chances of uncertainty that make the prediction more trustworthy.","PeriodicalId":125849,"journal":{"name":"2020 IEEE/ACM 13th International Conference on Utility and Cloud Computing (UCC)","volume":"38 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115319880","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 : 2020-12-01DOI: 10.1109/UCC48980.2020.00040
Pradyumna Kaushik, S. Raghavendra, M. Govindaraju, Devesh Tiwari
We propose a resource allocation policy that uses (a) Power as a first class parameter as an indicator of the computational intensity of a task and its potential impact on peak power draw, and (b) Power Tolerance as an indicator of a task’s sensitivity towards degradation of performance as a result of resource contention. Through experimentation and analysis, we present coarse-grained and fine-grained Power Tolerance assignment methods that can be employed to make smarter peak power performance trade-offs. Our experiments show that (a) cloud operators can benefit from a uniform workload-wide setting of Power Tolerance to achieve significant reduction in peak power consumption, (b) fine-grained Power Tolerance assignment methods show potential in making smarter peak power and performance trade-offs.
{"title":"Exploring the Potential of using Power as a First Class Parameter for Resource Allocation in Apache Mesos Managed Clouds","authors":"Pradyumna Kaushik, S. Raghavendra, M. Govindaraju, Devesh Tiwari","doi":"10.1109/UCC48980.2020.00040","DOIUrl":"https://doi.org/10.1109/UCC48980.2020.00040","url":null,"abstract":"We propose a resource allocation policy that uses (a) Power as a first class parameter as an indicator of the computational intensity of a task and its potential impact on peak power draw, and (b) Power Tolerance as an indicator of a task’s sensitivity towards degradation of performance as a result of resource contention. Through experimentation and analysis, we present coarse-grained and fine-grained Power Tolerance assignment methods that can be employed to make smarter peak power performance trade-offs. Our experiments show that (a) cloud operators can benefit from a uniform workload-wide setting of Power Tolerance to achieve significant reduction in peak power consumption, (b) fine-grained Power Tolerance assignment methods show potential in making smarter peak power and performance trade-offs.","PeriodicalId":125849,"journal":{"name":"2020 IEEE/ACM 13th International Conference on Utility and Cloud Computing (UCC)","volume":"2 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116789119","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 : 2020-12-01DOI: 10.1109/UCC48980.2020.00056
L. Larsson, H. Gustafsson, C. Klein, E. Elmroth
This position paper presents our vision for a distributed decentralized Kubernetes federation control plane. The goal is to support federations consisting of thousands of Kubernetes clusters, in order to support next generation edge cloud use-cases. Our review of the literature and experience with the current centralized state of the art Kubernetes federation controllers shows that it is unable to scale to a sufficient size, and centralization constitutes an unacceptable single point of failure. Our proposed system maintains cluster autonomy, allows clusters to collaboratively handle error conditions, and scales to support edge cloud use-cases. Our approach is based on a shared database of conflict-free replicated data types (CRDTs), shared among all clusters in the federation, and algorithms that make use of the data.
{"title":"Decentralized Kubernetes Federation Control Plane","authors":"L. Larsson, H. Gustafsson, C. Klein, E. Elmroth","doi":"10.1109/UCC48980.2020.00056","DOIUrl":"https://doi.org/10.1109/UCC48980.2020.00056","url":null,"abstract":"This position paper presents our vision for a distributed decentralized Kubernetes federation control plane. The goal is to support federations consisting of thousands of Kubernetes clusters, in order to support next generation edge cloud use-cases. Our review of the literature and experience with the current centralized state of the art Kubernetes federation controllers shows that it is unable to scale to a sufficient size, and centralization constitutes an unacceptable single point of failure. Our proposed system maintains cluster autonomy, allows clusters to collaboratively handle error conditions, and scales to support edge cloud use-cases. Our approach is based on a shared database of conflict-free replicated data types (CRDTs), shared among all clusters in the federation, and algorithms that make use of the data.","PeriodicalId":125849,"journal":{"name":"2020 IEEE/ACM 13th International Conference on Utility and Cloud Computing (UCC)","volume":"72 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116724103","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 : 2020-12-01DOI: 10.1109/ucc48980.2020.00018
{"title":"Keynotes [7 abstracts]","authors":"","doi":"10.1109/ucc48980.2020.00018","DOIUrl":"https://doi.org/10.1109/ucc48980.2020.00018","url":null,"abstract":"","PeriodicalId":125849,"journal":{"name":"2020 IEEE/ACM 13th International Conference on Utility and Cloud Computing (UCC)","volume":"43 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127879538","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 : 2020-12-01DOI: 10.1109/ucc48980.2020.00017
{"title":"Doctoral Symposium Technical Program Committee","authors":"","doi":"10.1109/ucc48980.2020.00017","DOIUrl":"https://doi.org/10.1109/ucc48980.2020.00017","url":null,"abstract":"","PeriodicalId":125849,"journal":{"name":"2020 IEEE/ACM 13th International Conference on Utility and Cloud Computing (UCC)","volume":"23 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134233322","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 : 2020-12-01DOI: 10.1109/UCC48980.2020.00044
Luke Lockhart, P. Harvey, Pierre Imai, P. Willis, B. Varghese
Partitioning and distributing deep neural networks (DNNs) across end-devices, edge resources and the cloud has a potential twofold advantage: preserving privacy of the input data, and reducing the ingress bandwidth demand beyond the edge. However, for a given DNN, identifying the optimal partition configuration for distributing the DNN that maximizes performance is a significant challenge. This is because the combination of potential target hardware resources that maximizes performance and the sequence of layers of the DNN that should be distributed across the target resources needs to be determined, while accounting for user-defined objectives/constraints for partitioning. This paper presents Scission, a tool for automated benchmarking of DNNs on a given set of target device, edge and cloud resources for determining optimal partitions that maximize DNN performance. The decision-making approach is context-aware by capitalizing on hardware capabilities of the target resources, their locality, the characteristics of DNN layers, and the network condition. Experimental studies are carried out on 18 DNNs. The decisions made by Scission cannot be manually made by a human given the complexity and the number of dimensions affecting the search space. The benchmarking overheads of Scission allow for responding to operational changes periodically rather than in real-time. Scission is available for public download 1.
{"title":"Scission: Performance-driven and Context-aware Cloud-Edge Distribution of Deep Neural Networks","authors":"Luke Lockhart, P. Harvey, Pierre Imai, P. Willis, B. Varghese","doi":"10.1109/UCC48980.2020.00044","DOIUrl":"https://doi.org/10.1109/UCC48980.2020.00044","url":null,"abstract":"Partitioning and distributing deep neural networks (DNNs) across end-devices, edge resources and the cloud has a potential twofold advantage: preserving privacy of the input data, and reducing the ingress bandwidth demand beyond the edge. However, for a given DNN, identifying the optimal partition configuration for distributing the DNN that maximizes performance is a significant challenge. This is because the combination of potential target hardware resources that maximizes performance and the sequence of layers of the DNN that should be distributed across the target resources needs to be determined, while accounting for user-defined objectives/constraints for partitioning. This paper presents Scission, a tool for automated benchmarking of DNNs on a given set of target device, edge and cloud resources for determining optimal partitions that maximize DNN performance. The decision-making approach is context-aware by capitalizing on hardware capabilities of the target resources, their locality, the characteristics of DNN layers, and the network condition. Experimental studies are carried out on 18 DNNs. The decisions made by Scission cannot be manually made by a human given the complexity and the number of dimensions affecting the search space. The benchmarking overheads of Scission allow for responding to operational changes periodically rather than in real-time. Scission is available for public download 1.","PeriodicalId":125849,"journal":{"name":"2020 IEEE/ACM 13th International Conference on Utility and Cloud Computing (UCC)","volume":"67 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125183268","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 : 2020-12-01DOI: 10.1109/UCC48980.2020.00020
Tuhin Chakraborty, A. Toosi, C. Kopp, Peter James Stuckey, Julien Mahet
In recent times, both industry and academia have paid significant attention to the power management of cloud data centers (CDCs), due to their typically very high electrical energy consumption. While servers remain the components with the highest power-consumption, network stacks can also consume about 10-20 percent of the total energy usage in a data center. Dynamic Virtual Machine (VM) consolidation is one way to reduce the number of active servers, which can be done by live migration of the VMs. But, migration operations in a data center bring several system and service level overheads that include downtime, elephant flows over the network, and potentially higher failure rates. In this work, we propose algorithms for minimizing the number of VM migrations to attain the optimized joint host-network power consumption in a cloud data center. We present a trade-off between the number of migrations, the joint host-network power consumption, and the computational time complexity of the proposed algorithms. Using Mininet and ONOS, an SDN enabled framework is utilised to evaluate the proposed algorithms. Experimental results show that our algorithms can reduce power consumption by about 11 percent, while completing between 18 to 25 percent more VM migrations compared to the baseline algorithm, which only minimizes migration without guaranteeing lowest power consumption.
{"title":"Joint Host-Network Power Scaling with Minimizing VM Migration in SDN-enabled Cloud Data Centers","authors":"Tuhin Chakraborty, A. Toosi, C. Kopp, Peter James Stuckey, Julien Mahet","doi":"10.1109/UCC48980.2020.00020","DOIUrl":"https://doi.org/10.1109/UCC48980.2020.00020","url":null,"abstract":"In recent times, both industry and academia have paid significant attention to the power management of cloud data centers (CDCs), due to their typically very high electrical energy consumption. While servers remain the components with the highest power-consumption, network stacks can also consume about 10-20 percent of the total energy usage in a data center. Dynamic Virtual Machine (VM) consolidation is one way to reduce the number of active servers, which can be done by live migration of the VMs. But, migration operations in a data center bring several system and service level overheads that include downtime, elephant flows over the network, and potentially higher failure rates. In this work, we propose algorithms for minimizing the number of VM migrations to attain the optimized joint host-network power consumption in a cloud data center. We present a trade-off between the number of migrations, the joint host-network power consumption, and the computational time complexity of the proposed algorithms. Using Mininet and ONOS, an SDN enabled framework is utilised to evaluate the proposed algorithms. Experimental results show that our algorithms can reduce power consumption by about 11 percent, while completing between 18 to 25 percent more VM migrations compared to the baseline algorithm, which only minimizes migration without guaranteeing lowest power consumption.","PeriodicalId":125849,"journal":{"name":"2020 IEEE/ACM 13th International Conference on Utility and Cloud Computing (UCC)","volume":"104 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117197417","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 : 2020-12-01DOI: 10.1109/UCC48980.2020.00037
M. Aslanpour, A. Toosi, R. Gaire, M. A. Cheema
Mechanisms for dynamically adding and removing Virtual Machines (VMs) to reduce cost while minimizing the latency are called auto-scaling. Latency improvements are mainly fulfilled through minimizing the "average" response times while unpredictabilities and fluctuations of the Web applications, aka flash crowds, can result in very high latencies for users’ requests. Requests influenced by flash crowd suffer from long latencies, known as outliers. Such outliers are inevitable to a large extent as auto-scaling solutions continue to improve the average, not the "tail" of latencies. In this paper, we study possible sources of tail latency in auto-scaling mechanisms for Web applications. Based on our extensive evaluations in a real cloud platform, we discovered sources of a tail latency as 1) large requests, i.e. those data-intensive; 2) long-term scaling intervals; 3) instant analysis of scaling parameters; 4) conservative, i.e. tight, threshold tuning; 5) load-unaware surplus VM selection policies used for executing a scale-down decision; 6) cooldown feature, although cost-effective; and 7) VM start-up delay. We also discovered that after improving the average latency by auto-scaling mechanisms, the tail may behave differently, demanding dedicated tail-aware solutions for auto-scaling mechanisms.
{"title":"Auto-scaling of Web Applications in Clouds: A Tail Latency Evaluation","authors":"M. Aslanpour, A. Toosi, R. Gaire, M. A. Cheema","doi":"10.1109/UCC48980.2020.00037","DOIUrl":"https://doi.org/10.1109/UCC48980.2020.00037","url":null,"abstract":"Mechanisms for dynamically adding and removing Virtual Machines (VMs) to reduce cost while minimizing the latency are called auto-scaling. Latency improvements are mainly fulfilled through minimizing the \"average\" response times while unpredictabilities and fluctuations of the Web applications, aka flash crowds, can result in very high latencies for users’ requests. Requests influenced by flash crowd suffer from long latencies, known as outliers. Such outliers are inevitable to a large extent as auto-scaling solutions continue to improve the average, not the \"tail\" of latencies. In this paper, we study possible sources of tail latency in auto-scaling mechanisms for Web applications. Based on our extensive evaluations in a real cloud platform, we discovered sources of a tail latency as 1) large requests, i.e. those data-intensive; 2) long-term scaling intervals; 3) instant analysis of scaling parameters; 4) conservative, i.e. tight, threshold tuning; 5) load-unaware surplus VM selection policies used for executing a scale-down decision; 6) cooldown feature, although cost-effective; and 7) VM start-up delay. We also discovered that after improving the average latency by auto-scaling mechanisms, the tail may behave differently, demanding dedicated tail-aware solutions for auto-scaling mechanisms.","PeriodicalId":125849,"journal":{"name":"2020 IEEE/ACM 13th International Conference on Utility and Cloud Computing (UCC)","volume":"28 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122662203","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 : 2020-12-01DOI: 10.1109/ucc48980.2020.00011
Shuihua Wang
Due to the proliferation of biomedical imaging modalities such as Photoacoustic Tomography, Computed Tomography (CT), etc., massive amounts of biomedical data are being generated on a daily basis. How can we utilize such big data to build better health profiles and better predictive models so that we can better diagnose and treat diseases and provide a better life for humans? In the past years, many successful learning methods such as deep learning were proposed to answer this crucial question, which has social, economic, as well as legal implications. However, several significant problems plague the processing of big biomedical data, such as data heterogeneity, data incompleteness, data imbalance, and high dimensionality. What is worse is that many data sets exhibit multiple such problems. Most existing learning methods can only deal with homogeneous, complete, class-balanced, and moderate-dimensional data. Therefore, data preprocessing techniques including data representation learning, dimensionality reduction, and missing value imputation should be developed to enhance the applicability of deep learning methods in real-world applications of biomedicine.
{"title":"Message from the B2D2LM 2020 Workshop Chairs","authors":"Shuihua Wang","doi":"10.1109/ucc48980.2020.00011","DOIUrl":"https://doi.org/10.1109/ucc48980.2020.00011","url":null,"abstract":"Due to the proliferation of biomedical imaging modalities such as Photoacoustic Tomography, Computed Tomography (CT), etc., massive amounts of biomedical data are being generated on a daily basis. How can we utilize such big data to build better health profiles and better predictive models so that we can better diagnose and treat diseases and provide a better life for humans? In the past years, many successful learning methods such as deep learning were proposed to answer this crucial question, which has social, economic, as well as legal implications. However, several significant problems plague the processing of big biomedical data, such as data heterogeneity, data incompleteness, data imbalance, and high dimensionality. What is worse is that many data sets exhibit multiple such problems. Most existing learning methods can only deal with homogeneous, complete, class-balanced, and moderate-dimensional data. Therefore, data preprocessing techniques including data representation learning, dimensionality reduction, and missing value imputation should be developed to enhance the applicability of deep learning methods in real-world applications of biomedicine.","PeriodicalId":125849,"journal":{"name":"2020 IEEE/ACM 13th International Conference on Utility and Cloud Computing (UCC)","volume":"54 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115930403","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：