In this paper, we propose a cost sensitive matrix factorization (CSMF) for face recognition. To make the face representation cost sensitive, CSMF adopts a more flexible feature embedding strategy. It contains two main...
详细信息
ISBN:
(数字)9783319689357
ISBN:
(纸本)9783319689357;9783319689340
In this paper, we propose a cost sensitive matrix factorization (CSMF) for face recognition. To make the face representation cost sensitive, CSMF adopts a more flexible feature embedding strategy. It contains two main steps: (1) matrix factorization for the learning of latent semantic representation and (2) cost sensitive latent semantic regression. In this way, the face images are embedded into their label space withthe misclassification loss minimized. the experimental results on Extended Yale B and ORL demonstrate its effectiveness.
In this paper we develop a previous work on matching data [2], inserting their contents in the more general framework of contingency tables and dealing withthe dimensions problem generated by the combination of the m...
详细信息
ISBN:
(纸本)9783319689357;9783319689340
In this paper we develop a previous work on matching data [2], inserting their contents in the more general framework of contingency tables and dealing withthe dimensions problem generated by the combination of the multiple characteristics that define each row and column category. Two concepts related to the matching process are defined: propensity to match and similarity in the matching. Both measures can be divided into partial components which allow a better understanding of the underlying structure of the data. We illustrate our methodology taking as an example a labor market where each worker category and each job category is defined by the combination of two attributes: location and occupational level.
this paper studies energy harvesting communication systems in which the base station sends data packets using the energy harvested from the surrounding natural environment. We assume that at each time slot only inform...
详细信息
ISBN:
(纸本)9781538605349;9781538605332
this paper studies energy harvesting communication systems in which the base station sends data packets using the energy harvested from the surrounding natural environment. We assume that at each time slot only information about the current and past state of the base station is available, modeling the scenario as a markov decision process and propose a reinforcement learning approach based on Q-learning for the transmitter to learn to cooperation trough energy sharing. For the problem of continuous state space in reinforcement learning, we propose an online algorithm based on approximate linear dependence to sparse sample pool. Furthermore, we use a neural network as approximator for the value function to improve the generalization ability. Numerical results show that our proposed algorithm can significantly improve the system performance even when the future environment change and energy harvested is uncertain for the BS.
Promoting data sharing between organisations is challenging, without the added concerns over having actions traced. Even with encrypted search capabilities, the entities digital location and downloaded information can...
详细信息
ISBN:
(纸本)9781509049066
Promoting data sharing between organisations is challenging, without the added concerns over having actions traced. Even with encrypted search capabilities, the entities digital location and downloaded information can be traced, leaking information to the hosting organisation. this is a problem for law enforcement and government agencies, where any information leakage is not acceptable, especially for investigations. Anonymous routing is a technique to stop a host learning which agency is accessing information. Many related works for anonymous routing have been proposed, but are designed for Internet traffic, and are over complicated for internal usage. A streaming design for circuit creation is proposed using elliptic curve cryptography. Allowing for a simple anonymous routing solution, which provides fast performance with source and destination anonymity to other organisations.
Clustering is a fundamental tool that has been applied in dealing with huge volumes of text documents and images. For extracting relevant information from the enormous volumes of available data, some co-clustering alg...
详细信息
ISBN:
(纸本)9783319689357;9783319689340
Clustering is a fundamental tool that has been applied in dealing with huge volumes of text documents and images. For extracting relevant information from the enormous volumes of available data, some co-clustering algorithms have been proposed and shown to be superior to traditional one-side clustering. In this paper, we proposed a novel co-clustering approach called double sparse manifold learning (DSML). We based our formulation on double sparse constraints and manifold learning which use a modified version of mutual k-nearest neighbor graph to capture the underlying structure, modeled sample-feature relationship from the data reconstruction perspective. We developed an iterative procedure to get the solution. Our method preserves local geometrical structure better. Experiments on three benchmark datasets show that our method can get more promising performance on all analyzed data-sets.
A major challenge in Infrastructure as a Service (IaaS) clouds is its exposure to malware. Malware can spread rapidly within a datacenter and can cause major disruption to a cloud service provider and its clients. thi...
详细信息
A major challenge in Infrastructure as a Service (IaaS) clouds is its exposure to malware. Malware can spread rapidly within a datacenter and can cause major disruption to a cloud service provider and its clients. this paper introduces and discusses an effective malware detection approach in cloud infrastructure using Convolutional Neural Network (CNN), a deep learning approach. We initially employ a standard 2d CNN by training on metadata available for each of the processes in a virtual machine (VM) obtained by means of the hypervisor. We enhance the CNN classifier accuracy by using a novel 3d CNN (where an input is a collection of samples over a time interval), which greatly helps reduce mislabelled samples during data collection and training. Our experiments are performed on data collected by running various malware (mostly Trojans and Rootkits) on VMs. the malware used in our experiments are randomly selected. this reduces the selection bias of known-to-be highly active malware for easy detection. We demonstrate that our 2d CNN model reaches an accuracy of ≃ 79%, and our 3d CNN model significantly improves the accuracy to ≃ 90%.
Paper describes an approach of deep learning for QRS wave detection for using in mobile heart monitoring systems. Authors analyze a deep learning approach and its advantages in the field of feature extraction and dete...
详细信息
ISBN:
(纸本)9781538605011
Paper describes an approach of deep learning for QRS wave detection for using in mobile heart monitoring systems. Authors analyze a deep learning approach and its advantages in the field of feature extraction and detection, and deep network architecture. Two different variants of deep network are proposed. ECG data processing scheme that includes a neural network is described. It presumes preprocessing, filtering, windowing of ECG signal, buffering, QRS wave detection and analysis. Network training process is mathematically founded. Two variants of neural network are experimentally tested. Training sets and test sets are obtained from free ECG data bank ***. Experimental results show that network with decreasing number of neurons in hidden layers has a better generalization capability. Next steps of research will include experiments with training set size and determining of its' influence on the quality of detection.
A novel discriminant locality preserving dictionary learning (DLPDL) algorithm for face recognition is proposed in this paper. In order to achieve better performance and less computation, dimensionality reduction is a...
详细信息
ISBN:
(纸本)9783319689357;9783319689340
A novel discriminant locality preserving dictionary learning (DLPDL) algorithm for face recognition is proposed in this paper. In order to achieve better performance and less computation, dimensionality reduction is applied on original image samples. Most of the proposed dictionary learning methods learn features and dictionary, however, the inner structure of feature is hardly considered. therefore, by incorporating discriminant locality preserving criteria into dictionary learning, the margin of coefficients distance between between-class and within-class is encourage to be large in order to enhance the classification ability and gain discriminative information. What is more, the local structure of the feature is also preserved, which is very vital in face recognition performance. Our experiments on Extended Yale B, AR and CMU face database demonstrated the proposed algorithm has higher recognition performance than other dictionary learning based classification methods.
Withthe development of data mining technology, lack of private resource protection has become a serious challenge. We propose to clarify the expression of Knowledge Graph in three layers including data Graph, Informa...
详细信息
ISBN:
(纸本)9783319689357;9783319689340
Withthe development of data mining technology, lack of private resource protection has become a serious challenge. We propose to clarify the expression of Knowledge Graph in three layers including data Graph, Information Graph and Knowledge Graph and illustrate the representation of data Graph, Information Graph and Knowledge Graph respectively. We elaborate a pay as you use resource security provision approach based on data Graph, Information Graph and Knowledge Graph in order to ensure that resources will not be used, tampered with, lost and destroyed in unauthorized situations.
this paper mainly proposes an parallel distributed learning algorithm for the robust convex optimization (RCO). Firstly, the scenario approach is used to transform RCO into its probabilistic approximation Scenario Pro...
详细信息
ISBN:
(纸本)9783319689357;9783319689340
this paper mainly proposes an parallel distributed learning algorithm for the robust convex optimization (RCO). Firstly, the scenario approach is used to transform RCO into its probabilistic approximation Scenario Problem (SP), which is distributively solved by multiprocessors to lighten the computational burden. Secondly, each processor (node) of the colored network processes the local optimization via a primal-dual subgradient algorithm (PDSA) to obtain an optimal solution called a local variable. Finally, a consensus method named the Colored Distributed Average Consensus (CDAC), which is based on Distributed Average Consensus (DAC), is proposed to act on the whole local variables to obtain the global optimal solution. Experimental results show that CDAC has an advantage in terms of computational time over DAC, while they have the same results.
暂无评论