Coordinated checkpointing is a widely-used checkpoint/restart protocol for fault-tolerance in large-scale HPC systems. However, this protocol will involve massive amounts of I/O concentration, resulting in considerably high checkpoint overhead and high energy consumption. This paper focuses on speculative checkpointing, a CPR mechanism that allows for temporal distribution of checkpointings to avoid I/O concentration. We propose execution time and energy models for speculative checkpointing, and investigate energy-performance characteristics when speculative checkpointing is adopted in exascale systems. Using these models, we study the benefit of speculative checkpointing over coordinated checkpointing under various realistic scenarios for exascale HPC systems. We show that, compared to coordinated checkpointing, speculative checkpointing can achieve up to a 11% energy reduction at the cost of a relatively-small increase in the execution time. In addition, a significant energy-performance trade-off is expected when the system scale exceeds 1.2 million nodes.

Existing image deblurring methods with a blurred/noisy image pair take a two-step approach: blur kernel estimation and image restoration. They can achieve better and much more stable blur kernel estimation than single image deblurring methods. On the other hand, in the image restoration step, they do not exploit the information on the noisy image, or they require ad hoc tuning of interdependent parameters. This paper focuses on the image restoration step and proposes a new restoration method of using a blurred/noisy image pair. In our method, the image restoration problem is formulated as a constrained convex optimization problem, where data-fidelity to a blurred image and that to a noisy image is properly taken into account as multiple hard constraints. This offers (i) high quality restoration when the blurred image also contains noise; (ii) robustness to the estimation error of the blur kernel; and (iii) easy parameter setting. We also provide an efficient algorithm for solving our optimization problem based on the so-called alternating direction method of multipliers (ADMM). Experimental results support our claims.

This paper presents a novel speaking aid system to help laryngectomees produce more naturally sounding electrolaryngeal (EL) speech. An electrolarynx is an external device to generate excitation signals, instead of vibration of the vocal folds. Although the conventional EL speech is quite intelligible, its naturalness suffers from the unnatural fundamental frequency (F0) patterns of the mechanically generated excitation signals. To improve the naturalness of EL speech, we have proposed EL speech enhancement methods using statistical F0 pattern prediction. In these methods, the original EL speech recorded by a microphone is presented from a loudspeaker after performing the speech enhancement. These methods are effective for some situation, such as telecommunication, but it is not suitable for face-to-face conversation because not only the enhanced EL speech but also the original EL speech is presented to listeners. In this paper, to develop an EL speech enhancement also effective for face-to-face conversation, we propose a method for directly controlling F0 patterns of the excitation signals to be generated from the electrolarynx using the statistical F0 prediction. To get an "actual feel” of the proposed system, we also implement a prototype system. By using the prototype system, we find latency issues caused by a real-time processing. To address these latency issues, we furthermore propose segmental continuous F0 pattern modeling and forthcoming F0 pattern modeling. With evaluations through simulation, we demonstrate that our proposed system is capable of effectively addressing the issues of latency and those of electrolarynx in term of the naturalness.

This paper proposes a novel framework for enhancing underwater images captured by optical imaging model and non-local means denoising. The proposed approach adjusts the color balance using biasness correction and the average luminance. Scene visibility is then enhanced based on an underwater optical imaging model. The increase in noise in the enhanced images is alleviated by non-local means (NLM) denoising. The final enhanced images are characterized by improved visibility while retaining color fidelity and reducing noise. The proposed method does not require specialized hardware nor prior knowledge of the underwater environment.

In the field of action recognition, Spatio-Temporal Interest Points (STIPs)-based features have shown high efficiency and robustness. However, most of state-of-the-art work to describe STIPs, they typically focus on 2-dimensions (2D) images, which ignore information in 3D spatio-temporal space. Besides, the compact representation of descriptors should be considered due to the costs of storage and computational time. In this paper, a novel local descriptor named 3D Gradient LBP is proposed, which extends the traditional descriptor Local Binary Patterns (LBP) into 3D spatio-temporal space. The proposed descriptor takes advantage of the neighbourhood information of cuboids in three dimensions, which accounts for its excellent descriptive power for the distribution of grey-level space. Experiments on three challenging datasets (KTH, Weizmann and UT Interaction) validate the effectiveness of our approach in the recognition of human actions.

Restoration is an important area in improving the visual quality, and lays the foundation for accurate object detection or terrain classification in image analysis. In this paper, we introduce Beta process priors into hierarchical sparse Bayesian learning for recovering underlying degraded hyperspectral images (HSI), including suppressing the various noises and inferring the missing data. The proposed method decomposes the HSI into the weighted summation of the dictionary elements, Gaussian noise term and sparse noise term. With these, the latent information and the noise characteristics of HSI can be well learned and represented. Solved by Gibbs sampler, the underlying dictionary and the noise can be efficiently predicted with no tuning of any parameters. The performance of the proposed method is compared with state-of-the-art ones and validated on two hyperspectral datasets, which are contaminated with the Gaussian noises, impulse noises, stripes and dead pixel lines, or with a large number of data missing uniformly at random. The visual and quantitative results demonstrate the superiority of the proposed method.

We propose an appearance-based proficiency evaluation methodology based on fine-motion analysis. We consider the effects of individual habit in evaluating proficiency and analyze the fine motion of guitar-picking. We first extract multiple features on a large number of dense trajectories of fine motion. To facilitate analysis, we then generate a histogram of motion features using a bag-of-words model and change the number of visual words as appropriate. To remove the effects of individual habit, we extract the common principal histogram elements corresponding to experts or beginners according to discrimination's contribution rates using random forests. We finally calculate the similarity of the histograms to evaluate the proficiency of a guitar-picking motion. By optimizing the number of visual words for proficiency evaluation, we demonstrate that our method distinguishes experts from beginners with an accuracy of about 86%. Moreover, we verify experimentally that our proposed methodology can evaluate proficiency while removing the effects of individual habit.

Speaker verification is the task of determining whether two utterances represent the same person. After representing the utterances in the i-vector space, the crucial problem is only how to compute the similarity of two i-vectors. Metric learning has provided a viable solution to this problem. Until now, many metric learning algorithms have been proposed, but they are usually limited to learning a linear transformation. In this paper, we propose a nonlinear metric learning method, which learns an explicit mapping from the original space to an optimal subspace using deep Restricted Boltzmann Machine network. The proposed method is evaluated on the NIST SRE 2008 dataset. Since the proposed method has a deep learning architecture, the evaluation results show superior performance than some state-of-the-art methods.

Content-centric networking (CCN) is an emerging networking architecture that is being actively investigated in both the research and industrial communities. In the latest version of CCN, a large number of interests have to be issued when large content is retrieved. Since CCN routers have to search several tables for each incoming interest, this could cause a serious problem of router workload. In order to solve this problem, this paper introduces a novel strategy of “grouping” multiple interests with common information and “packing” them to a special interest called the list interest. Our list interest is designed to co-operate with the manifest of CCN as its dual. This paper demonstrates that by skipping and terminating several search steps using the common information in the list interest, the router can search its tables for the list interest-based request with dramatically smaller complexity than the case of the standard interest-based request. Furthermore, we also consider the deployment of list interests and design a novel TCP-like congestion control method for list interests to employ them just like standard interests.

Hough-based voting approaches have been widely used to solve many detection problems such as object and action detection. These approaches for action detection cast votes for action classes and positions based on the local spatio-temporal features of given videos. The voting process of each local feature is performed independently of the other local features. This independence enables the method to be robust to occlusions because votes based on visible local features are not influenced by occluded local features. However, such independence makes discrimination of similar motions between different classes difficult and causes the method to cast many false votes. We propose a novel Hough-based action detection method to overcome the problem of false votes. The false votes do not occur randomly such that they depend on relevant action classes. We introduce vote distributions, which represent the number of votes for each action class. We assume that the distribution of false votes include important information necessary to improving action detection. These distributions are used to build a model that represents the characteristics of Hough voting that include false votes. The method estimates the likelihood using the model and reduces the influence of false votes. In experiments, we confirmed that the proposed method reduces false positive detection and improves action detection accuracy when using the IXMAS dataset and the UT-Interaction dataset.

In satellite/terrestrial integrated mobile communication systems (STICSs), a user terminal directly connects both terrestrial and satellite base stations. STICS enables expansion of service areas and provides a robust communication service for large disasters. However, the cell radius of the satellite system is large (approximately 100km), and thus a capacity enhancement of the satellite subsystem for accommodating many users is needed. Therefore, in this paper, we propose an application of two methods — multiple-input multiple-output (MIMO) transmission using multi-satellites and non-orthogonal multiple access (NOMA) for STICS — to realize the performance improvement in terms of system capacity and user fairness. Through numerical simulations, we show that system capacity and user fairness are increased by the proposed scheme that applies the two methods.

In this paper, we first analyze the discriminative power in the Best Match (BM) 25 formula and provide its calculation method from the Bayesian point of view. The resulting, derived discriminative power is quite similar to the exponential inverse document frequency (EIDF) that we have previously proposed [1] but retains more preferable theoretical advantages. In our previous paper [1], we proposed the EIDF in the framework of the probabilistic information retrieval (IR) method BM25 to address the instance search task, which is a specific object search for videos using an image query. Although the effectiveness of our EIDF was experimentally demonstrated, we did not consider its theoretical justification and interpretation. We also did not describe the use of region-of-interest (ROI) information, which is supposed to be input to the instance search system together with the original image query showing the instance. Therefore, here, we justify the EIDF by calculating the discriminative power in the BM25 from the Bayesian viewpoint. We also investigate the effect of the ROI information for improving the instance search accuracy and propose two search methods incorporating the ROI effect into the BM25 video ranking function. We validated the proposed methods through a series of experiments using the TREC Video Retrieval Evaluation instance search task dataset.

In this research, we investigated the reliability of a 1-out-of-2 system with two-stage repair comprising hardware restoration and data reconstruction modes. Hardware restoration is normally independently executed by two modules. In contrast, we assumed that one of the modules could omit data reconstruction by replicating the data from the module during normal operation. In this 1-out-of-2 system, the two modules mutually cooperated in the recovery mode. As a first step, an evaluation model using Markov chains was constructed to derive a reliability measure: “unavailability in steady state.” Numerical examples confirmed that the reliability of the system was improved by the use of two cooperating modules. As the data reconstruction time increased, the gains in terms of system reliability also increased.

This paper addresses the issue of implementing a sequence for restoring fiber links and communication paths that have failed due to a catastrophe. We present a mathematical formulation to minimize the total number of steps needed to restore communication paths. We also propose two heuristic algorithms: Minimum spanning tree - based degree order restoration and Congestion link order restoration. Numerical evaluations show that integer linear programming based order restoration yields the fewest number of restoration steps, and that the proposed heuristic algorithms, when used properly with regard to the accommodation rate, are highly effective for real-world networks.

In this letter, a local characteristic image restoration based on convolutional neural network is proposed. In this method, image restoration is considered as a classification problem and images are divided into several sub-blocks. The convolutional neural network is used to extract and classify the local characteristics of image sub-blocks, and the different forms of the regularization constraints are adopted for the different local characteristics. Experiments show that the image restoration results by the regularization method based on local characteristics are superior to those by the traditional regularization methods and this method also has lower computing cost.

In this paper we propose a novel classification method for the multiple k-nearest neighbor (MkNN) classifier and show its practical application to medical image processing. The proposed method performs fine classification when a pair of the spatial coordinate of the observation data in the observation space and its corresponding feature vector in the feature space is provided. The proposed MkNN classifier uses the continuity of the distribution of features of the same class not only in the feature space but also in the observation space. In order to validate the performance of the present method, it is applied to the tissue characterization problem of coronary plaque. The quantitative and qualitative validity of the proposed MkNN classifier have been confirmed by actual experiments.

Recovery of low-rank matrices has seen significant activity in many areas of science and engineering, motivated by theoretical results for exact reconstruction guarantees and interesting practical applications. Recently, numerous methods incorporated the nuclear norm to pursue the convexity of the optimization. However, this greatly restricts its capability and flexibility in dealing with many practical problems, where the singular values have clear physical meanings. This paper studies a generalized non-convex low-rank approximation, where the singular values are in lp-heuristic. Then specific results are derived for image restoration, including denoising and deblurring. Extensive experimental results on natural images demonstrate the improvement of the proposed method over the recent image restoration methods.

Local spatio-temporal features are popular in the human action recognition task. In practice, they are usually coupled with a feature encoding approach, which helps to obtain the video-level vector representations that can be used in learning and recognition. In this paper, we present an efficient local feature encoding approach, which is called Approximate Sparse Coding (ASC). ASC computes the sparse codes for a large collection of prototype local feature descriptors in the off-line learning phase using Sparse Coding (SC) and look up the nearest prototype's precomputed sparse code for each to-be-encoded local feature in the encoding phase using Approximate Nearest Neighbour (ANN) search. It shares the low dimensionality of SC and the high speed of ANN, which are both desired properties for a local feature encoding approach. ASC has been excessively evaluated on the KTH dataset and the HMDB51 dataset. We confirmed that it is able to encode large quantity of local video features into discriminative low dimensional representations efficiently.

Many methods have been proposed for measuring the structural similarity between two protein folds. However, it is difficult to select one best method from them for the classification task, as each method has its own strength and weakness. Intuitively, combining multiple methods is one solution to get the optimal classification results. In this paper, by generalizing the concept of the large margin nearest neighbor (LMNN), a method for combining multiple distance metrics from different types of protein structure comparison methods for protein fold classification task is proposed. While LMNN is limited to Mahalanobis-based distance metric learning from a set of feature vectors of training data, the proposed method learns an optimal combination of metrics from a set of distance metrics by minimizing the distances between intra-class data and enlarging the distances of different classes' data. The main advantage of the proposed method is the capability in finding an optimal weight coefficient for combination of many metrics, possibly including poor metrics, avoiding the difficulties in selecting which metrics to be included for the combination. The effectiveness of the proposed method is demonstrated on classification experiments using two public protein datasets, namely, Ding Dubchak dataset and ENZYMES dataset.

Nearest-neighbor-search classifiers are attractive but they have high intrinsic computational demands which limit their practical application. In this paper, we propose a coprocessor for k (k with k≥1) nearest neighbor (kNN) classification in which squared Euclidean distances (SEDs) are mapped into the clock domain for realizing high search speed and energy efficiency. The minimal SED searching is carried out by weighted frequency dividers that drastically reduce the normally exponential increase of the worst-case search-clock number with the bit width of vector components to only a linear increase. This also results in low power dissipation and high area-efficiency in comparison to the traditional method using large numbers of adders and comparators. The kNN classifier determines the class of an unknown input sample with a majority decision among the k nearest reference samples. The required majority-decision circuit is integrated with the clock-mapping-based minimal-SED searching architecture and proceeds with the classification immediately after identification of each of the k nearest references. A test chip in 180 nm CMOS technology, which can process 8 dimensions of 32 reference vectors in parallel, achieves low power dissipation of 40.32 mW (at 51.21 MHz clock frequency and 1.8 V supply voltage). Significantly, the distance search circuit consumes only 5.99 mW. Feature vectors with different dimensionality up to 2048 dimensions can be handled by the designed coprocessor due to a dimension extension circuit, enabling large flexibility for usage in different application.