Insufficient Memory Research Articles

3D Reconstruction system for collaborative scanning based on multiple RGB-D cameras

Current 3D reconstruction systems can easily fall into the problems of insufficient memory and low efficiency when reconstructing complex large scenes, and the reconstructed model drifts a lot at the same time. To solve the above problems, a 3D reconstruction system for collaborative scanning based on multiple RGB-D Cameras (Xtion sensors) is presented. In the case of no pose estimation, this paper can obtain the camera pose initial values of the sequence images through the image acquisition platform that has been calibrated in advance, which can well cope with the problems existing in the current 3D reconstruction system. The main innovations of this paper are as follows: (1) In view of the large amount of human-computer interaction required for the current 3D reconstruction system and the high requirements for data acquisition, a collaborative scanning 3D reconstruction system based on multiple Xiton sensors (RGB-D camera) is developed. The proposed system does not require human-computer interaction and can capture high quality image sequences fully automatically; (2) In view of the problems of memory shortage and low efficiency in current 3D reconstruction systems applied to complex large-scale scenes, this paper proposes a system which can obtain camera poses in advance calibration without the need for camera pose estimation; (3) In terms of camera pose optimization, a segmented bundle adjustment method has been presented to obtain the high-precision camera poses. A large number of experiments demonstrate that the proposed system can effectively solve the existing problems. Meanwhile, the proposed reconstruction system can obtain high-precision 3D models suitable for various complex large scenes, which can be widely used in human-computer interaction, virtual reality and other fields.

Efficient preference clustering via random fourier features

Approximations based on random Fourier features have recently emerged as an efficient and elegant method for designing large-scale machine learning tasks. Unlike approaches using the Nystrom method, which randomly samples the training examples, we make use of random Fourier features, whose basis functions (i.e., cosine and sine ) are sampled from a distribution independent from the training sample set, to cluster preference data which appears extensively in recommender systems. Firstly, we propose a two-stage preference clustering framework. In this framework, we make use of random Fourier features to map the preference matrix into the feature matrix, soon afterwards, utilize the traditional k-means approach to cluster preference data in the transformed feature space. Compared with traditional preference clustering, our method solves the problem of insufficient memory and greatly improves the efficiency of the operation. Experiments on movie data sets containing 100 000 ratings, show that the proposed method is more effective in clustering accuracy than the Nystrom and k-means, while also achieving better performance than these clustering approaches.

CT-image of rock samples super resolution using 3D convolutional neural network

Computed Tomography (CT) imaging technique is widely used in geological exploration, medical diagnosis and other fields. However, the resolution of CT images is usually limited by scanning devices and cost. Super-resolution (SR) methods based on deep learning provide remarkable performance for two-dimensional (2D) images. Unfortunately, few effective SR algorithms are available for three-dimensional (3D) images. This study proposes a novel network named as three-dimensional super-resolution convolutional neural network (3DSRCNN) to realize voxel SR imaging of rock samples. To solve the practical problems faced in the training process, such as slow convergence of network training and insufficient memory, we utilized adjustable learning rate, residual-learning, gradient clipping, momentum stochastic gradient descent (SGD) strategy to optimize training procedure. In addition, we have explored the empirical guidelines to set an appropriate number of network layers. Previous learning-based algorithms need to separately train samples for different scale factors; by contrast, our single model can perform the multi-scale SR. Further, our proposed method provides better performance in terms of PSNR, SSIM and efficiency compared with conventional methods.

Selective Mobile Cloud Offloading to Augment Multi-Persona Performance and Viability

Fueled by changes in professional application models, personal interests and desires and technological advances in mobile devices, multi-persona has emerged recently to keep balance between different aspects, in our daily life, on a single mobile terminal. In this context, mobile virtualization technology has turned the corner and currently heading towards widespread adoption to realize multi-persona. Although recent lightweight virtualization techniques were able to maintain balance between security and scalability of personas, the limited CPU power and insufficient memory and battery capacities, still threaten personas performance and viability. Throughout the last few years, cloud computing has cultivated and refined the concept of outsourcing computing resources, and nowadays, in the coming age of smartphones and tablets, the prerequisites are met for importing cloud computing to support resource constrained mobiles. From these premises, we propose in this paper a novel offloading-based approach that based on global resource usage monitoring, generic and adaptable problem formulation and heuristic decision making, is capable of augmenting personas performance and viability on mobile terminals. The experiments show its capability of reducing the resource usage overhead and energy consumption of the applications running in each persona, accelerating their execution and improving their scalability, allowing better adoption of multi-persona solution.

Intranasal immunization with pneumococcal surface protein A in the presence of nanoparticle forming polysorbitol transporter adjuvant induces protective immunity against the Streptococcus pneumoniae infection

Developing effective mucosal subunit vaccine for the Streptococcus pneumoniae has been unsuccessful mainly because of their poor immunogenicity with insufficient memory T and B cell responses. We thus address whether such limitation can be overcome by introducing effective adjuvants that can enhance immunity and show here that polysorbitol transporter (PST) serves as a mucosal adjuvant for a subunit vaccine against the Streptococcus pneumoniae. Pneumococcal surface protein A (PspA) with PST adjuvant induced protective immunity against S. pneumoniae challenge, especially long-term T and B cell immune responses. Moreover, we found that the PST preferentially induced T helper (Th) responses toward Th2 or T follicular helper (Tfh) cells and, importantly, that the responses were mediated through antigen-presenting cells via activating a peroxisome proliferator-activated receptor gamma (PPAR-γ) pathway. Thus, these data indicate that PST can be used as an effective and safe mucosal vaccine adjuvant against S. pneumoniae infection. State of SignificanceIn this study, we suggested the nanoparticle forming adjuvant, PST works as an effective adjuvant for the pneumococcal vaccine, PspA. The PspA subunit vaccine together with PST adjuvant efficiently induced protective immunity, even in the long-term memory responses, against Streptococcus pneumoniae lethal challenge. We found that PspA with PST adjuvant induced dendritic cell activation followed by follicular helper T cell responses through PPAR-γ pathway resulting long-term memory antibody-producing cells. Consequently, in this paper, we suggest the mechanism for safe nanoparticle forming subunit vaccine adjuvant against pneumococcal infection.

Parana: A Parallel Neural Architecture Considering Thermal Problem of 3D Stacked Memory

Recent advances in deep learning (DL) have stimulated increasing interests in neural networks (NN). From the perspective of operation type and network architecture, deep neural networks can be categorized into full convolution-based neural network (ConvNet), recurrent neural network (RNN), and fully-connected neural network (FCNet). Different types of neural networks are usually cascaded and combined as a hybrid neural network (Hybrid-NN) to complete real-life cognitive tasks. Such hybrid-NN implementation is memory-intensive with large number of memory accesses, hence the performance of hybrid-NN is often limited by the insufficient memory bandwidth. A “3D + 2.5D” integration system, which integrates a high-bandwidth 3D stacked DRAM side-by-side with a highly-parallel neural processing unit (NPU) on a silicon interposer, overcomes the bandwidth bottleneck in hybrid-NN acceleration. However, intensive concurrent 3D DRAM accesses produced by the NPU lead to a serious thermal problem in 3D DRAM. In this paper, we propose a neural processor called Parana for hybrid-NN acceleration in consideration of thermal problem of 3D DRAM. Parana solves the thermal problem of 3D memory by optimizing both the total number of memory accesses and memory accessing behaviors. For memory accessing behaviors, Parana balances the memory bandwidth by spatial division mapping hybrid-NN onto computing resources, which efficiently avoids that masses of memory accesses are issued in a short time period. To reduce the total number of memory accesses, we design a new NPU architecture and propose a memory-oriented tiling and scheduling mechanism to exploit the maximum utilization of on-chip buffer. Experimental results show that Parana reduces the peak temperature by up to 54.72 $^\circ$ C and the steady temperature by up to 32.27 $^\circ$ C over state-of-the-art accelerators with 3D memory without performance degradation.

Performance of Memory Virtualization Using Global Memory Resource Balancing

Virtualization has become a universal generalization layer in contemporary data centers. By multiplexing hardware resources into multiple virtual machines and facilitating several operating systems to run on the same physical platform at the same time, it can effectively decrease power consumption and building size or improve security by isolating virtual machines. In a virtualized system, memory resource supervision acts as a decisive task in achieving high resource employment and performance. Insufficient memory allocation to a virtual machine will degrade its performance drastically. On the contrasting, over allocation reasons ravage of memory resources. In the meantime, a virtual machine's memory stipulates may differ drastically. As a consequence, effective memory resource management calls for a dynamic memory balancer, which, preferably, can alter memory allocation in a timely mode for each virtual machine-based on their present memory stipulate and therefore realize the preeminent memory utilization and the best possible overall performance. Migrating operating system instances across discrete physical hosts is a helpful tool for administrators of data centers and clusters: It permits a clean separation among hardware and software, and make easy fault management. In order to approximate the memory, the stipulate of each virtual machine and to adjudicate probable memory resource disagreement, an extensively planned approach is to build an Least Recently Used based miss ratio curve which provides not only the current working set size but also the correlation between performance and the target memory allocation size. In this paper, the authors initially present a low overhead LRU-based memory demand tracking scheme, which includes three orthogonal optimizations: AVL based Least Recently Used association, dynamic hot set sizing. This assessment outcome confirms that, for the complete SPEC CPU 2006 benchmark set, subsequent to pertaining the 3 optimizing techniques, the mean overhead of MRC construction are lowered from 173% to only 2%. Based on current WSS, the authors then predict its trend in the near future and take different tactics for different forecast results. When there is an adequate amount of physical memory on the host, it locally balances its memory resource for the VMs. Once the local memory resource is insufficient and the memory pressure is predicted to sustain for a sufficiently long time, VM live migration, is used to move one or more VMs from the hot host to other host(s). Finally, for transient memory pressure, a remote cache is used to alleviate the temporary performance penalty. These experimental results show that this design achieves 49% center-wide speedup.

Hand-Held 3-D Reconstruction of Large-Scale Scene With Kinect Sensors Based on Surfel and Video Sequences

This letter presents a hand-held complex large-scale scene reconstruction method with Kinect sensors based on surfel and video sequences. The feature point method simultaneous localization and mapping (SLAM) is employed to estimate the pose of the camera, and then bundle adjustment by combining 2-D and 3-D feature points is used to optimize camera pose. Also, the surfel model is employed to construct deformation maps for the fusion and optimization of point clouds, and finally, an accurate precise 3-D map can be obtained. The main contribution of this letter is that: 1) by using the SLAM method to obtain camera pose as the initial value of optimization, the problem of insufficient memory and low efficiency of the structure form motion method can be well solved; 2) sparsely textured regions can be reconstructed better by using bundle adjustment by combining 2-D and 3-D feature points; and 3) dense 3-D reconstruction of large scenes can be achieved, and the reconstructed 3-D models are more elaborate. Finally, experimental results show that this proposed method can be applied to a variety of complex large-scale scenes, and can obtain accurate precise 3-D model. This presented 3-D reconstruction method can be widely used in the fields of human–computer interaction, consumer electronics, and virtual reality.

An Overview of Watershed Algorithm Implementations in Open Source Libraries

Watershed is a widespread technique for image segmentation. Many researchers apply the method implemented in open source libraries without a deep understanding of its characteristics and limitations. In the review, we describe benchmarking outcomes of six open-source marker-controlled watershed implementations for the segmentation of 2D and 3D images. Even though the considered solutions are based on the same algorithm by flooding having O(n)computational complexity, these implementations have significantly different performance. In addition, building of watershed lines grows processing time. High memory consumption is one more bottleneck for dealing with huge volumetric images. Sometimes, the usage of more optimal software is capable of mitigating the issues with the long processing time and insufficient memory space. We assume parallel processing is capable of overcoming the current limitations. However, the development of concurrent approaches for the watershed segmentation remains a challenging problem.

Real-time Big Data Processing System to Improve Semiconductor Production Efficiency in Smart Factory

Semiconductor production efficiency is closely related to the defect rate in the production process. The temperature and humidity control in the production line are very important because these affect the defect rate. So many smart factory of semiconductor production uses sensor. It is installed in the semiconductor process, which send huge amounts of data per second to a central server to carry out temperature and humidity control in each production line. However, big data processing systems that analyze and process large-scale data are subject to frequent delays in processing, and transmitted data are lost owing to bottlenecks and insufficient memory caused by traffic concentrated in the central server. In this paper, we propose a real-time big data processing system to improve semiconductor production efficiency. The proposed system consists of a production line collection system, task processing system and data storage system, and improves the productivity of the semiconductor manufacturing process by reducing data processing delays as well as data loss and discarded data.

3D Modeling Method of Whole Structure for Turbo-generator Based on PROE-ANSOFT Information Interaction

In this paper, a new method based on PROE-ANSOFT Information Interaction for building 3D model of turbo-generator structure is proposed, which can solve those problems such as low efficiency of surface modeling, the lack of physical connection relation for the automatic turbo-generator integral structure, the occurrence of insufficient memory and even crashes for imported PROE model. Taking the stator winding coil of a turbo-generator as an example, the establishment process of 3D solid model is given. On this basis, the overall structure of turbo-generator in the ANSOFT working platform is completed. And the advantages and disadvantages of five kinds of simulation models in ANSOFT working platform are compared and analyzed. The results show that the 3D simulation model drawn by this method can be no longer limited to the special solution environment of electromagnetic field symmetry. Meanwhile, this model can simulate the real physical connection relationship, and complete the division of the simulation model, which will be helpful for the further simulation analysis of the finite element software.

Tensor network subspace identification of polynomial state space models

This article introduces a tensor network subspace algorithm for the identification of specific polynomial state space models. The polynomial nonlinearity in the state space model is completely written in terms of a tensor network, thus avoiding the curse of dimensionality. We also prove how the block Hankel data matrices in the subspace method can be exactly represented by low rank tensor networks, reducing the computational and storage complexity significantly. The performance and accuracy of our subspace identification algorithm are illustrated by experiments, showing that our tensor network implementation identifies a seventh degree polynomial state space model around 20 times faster than the standard matrix implementation before the latter fails due to insufficient memory. The proposed algorithm is also robust with respect to noise and therefore applicable to practical systems.

Nie ma takiej wiedzy, która jest ostateczna. Każda stanowi przedsionek prowadzący do nowych obszarów niewiedzy. Götz i Meyer Davida Albahariego

This article’ s aim is to analyze David Albahari ’ s Gotz and Meyer as the only Serbian literary text whose main theme is the extermination of the native Jews. The recurring key-word of the analysis is the notion of void, which appears both on the plot level and a more metaphorical one that connotes a less perceptible meaning of Albahari ’ s prose which can be summarized as a conviction about the insufficient memory of the Holocaust in Serbia. The main aim of this paper is to demonstrate what method Albahari employs to build his story. Consequently, juxtaposing history (ostensibly objective, linked with the order of reason) with fiction proves to be here of the utmost importance. It demonstrates that this text – indirectly – inscribes itself in the debate on the possibilities of expressing Holocaust.

Application of Improved Eclat Algorithm in Students’ Evaluation of Teaching

The evaluation system of students is to find a way to solve the status way according to the exact needs of students and the teaching requirements of teachers, so as to improve the teaching level of teachers and improve the quality of school education. This paper uses the real evaluation sample and uses the data mining association rule algorithm to comprehensively analyze the massive data of the evaluation data and the basic information of the teacher. The purpose is to obtain the association rules between the teacher’s comprehensive information and its evaluation results. Using the evaluation data to explore its core issues. In this paper, the Eclat algorithm of association rules improves the problem of insufficient memory and occupying a large amount of time when searching for frequent itemsets in the data. The breadth-first algorithm is added to save operation time and improve the efficiency of the algorithm. The effectiveness of the improved algorithm is verified by comparative experiments and applied to the evaluation system so as to provide suggestions for the professional development of teachers from an objective perspective, and to build a harmonious, "people-oriented" evaluation system for students.

Distance metric optimization driven convolutional neural network for age invariant face recognition

Despite the great advances in face-related works in recent years, face recognition across age remains a challenging problem. The traditional approaches to this problem usually include two basic steps: feature extraction and the application of a distance metric, sometimes common space projection is also involved. On the one hand, handling these steps separately ignores the interactions of these components, and on the other hand, the fixed-distance threshold of measurement affects the model’s robustness. In this paper, we present a novel distance metric optimization driven learning approach that integrates these traditional steps via a deep convolutional neural network, which learns feature representations and the decision function in an end-to-end way. Given the labelled training images, we first generate a large number of pairs with a certain proportion of matched and unmatched pairs. For matched pairs, we try to select as many different age instances as possible for each person to learn the identification information that is not affected by age. Then, taking these pairs as input, we aim to enlarge the differences between the unmatched pairs while reducing the variations between the matched pairs, and we update the model parameters by using the mini-batch stochastic gradient descent (SGD) algorithm. Specifically, the distance matrix is used as the top fully connected layer, and the bottom layers representing the image features are integrated with it seamlessly. Thus, the image features and the distance metric can be optimized simultaneously by backward propagation. In particular, we introduce several training strategies to reduce the computational cost and overcome insufficient memory capacity. We evaluate our method on three tasks: age-invariant face identification on the MORPH database, age-invariant face retrieval on the CACD database and age-invariant face verification on CACD-VS database. The experimental results demonstrate the effectiveness of our approach.

Evaluation of triangular mesh layout techniques using large mesh simplification

Highly detailed polygonal meshes nowadays were created vastly from a great number of multimedia applications. Insufficient main memory space and inferior locality-of-reference nature of the input mesh sequence incurred frequent page swaps and extremely low runtime efficiency. To address this issue, techniques based on the models of graph layout, matrix bandwidth minimization, and space-filling-curves were reported. However, such models merely considered the optimization of vertex layout on the basis of the graph distance or matrix bandwidths while leave the optimization of face layout either unoptimized or resorted an additional sorting procedure. In this paper, we propose an improved theoretic model suggesting the optimization of index-faced triangular mesh layout on the basis of a jointly consideration on both the vertex and face layout. Unlike previous attempts, we do not need additional efforts in optimizing one layout after the other by sorting, the optimized layouts are generated on the fly with each other with respect to both the vertex or triangle bandwidths. According to the experimental results, our approach outperforms the CM and SFC methods not only at yielding better theoretical results but also at improving the runtime efficiency of large mesh simplification.

Memory-efficient algorithm for stored projection and backprojection matrix in helical CT.

Iterative image reconstruction is often time-consuming, especially for helical CT. The calculation of X-ray projections and backprojections are computationally expensive. Although they can be significantly accelerated by parallel computing (e.g., via graphics processing unit (GPU)), they have to be calculated numerous times on-the-fly (OTF) during iterative image reconstruction due to insufficient memory storage. In this work, the memory-efficient algorithm for stored system matrix (SSM) is developed for both projections and backprojections to avoid repeated OTF computations of system matrices. The SSM algorithm is based on the shift-invariance for projection and backprojection under a rotating coordinate. As a result, the size of projection and backprojection matrices can be significantly reduced and fully stored in memory. The proposed method can be readily incorporated into iterative reconstruction algorithm with minor modification, i.e., by replacing OTF for SSM. Rigorous mathematical analysis is carried out to establish the shift-invariance for ray-driven projection and pixel-driven backprojection. Numerical results via GPU suggest that the proposed SSM method has improved computational efficiency from the OTF method, i.e., by three- to sixfold acceleration for the projection and 3- to 16-fold acceleration for the backprojection respectively for helical CT. We propose a memory-efficient SSM algorithm for projections and backprojections so that system matrices can be fully stored on the state-of-the-art GPU to facilitate the rapid iterative helical CT image reconstruction.

Memory-Efficient Analysis of Dense Functional Connectomes.

The functioning of the human brain relies on the interplay and integration of numerous individual units within a complex network. To identify network configurations characteristic of specific cognitive tasks or mental illnesses, functional connectomes can be constructed based on the assessment of synchronous fMRI activity at separate brain sites, and then analyzed using graph-theoretical concepts. In most previous studies, relatively coarse parcellations of the brain were used to define regions as graphical nodes. Such parcellated connectomes are highly dependent on parcellation quality because regional and functional boundaries need to be relatively consistent for the results to be interpretable. In contrast, dense connectomes are not subject to this limitation, since the parcellation inherent to the data is used to define graphical nodes, also allowing for a more detailed spatial mapping of connectivity patterns. However, dense connectomes are associated with considerable computational demands in terms of both time and memory requirements. The memory required to explicitly store dense connectomes in main memory can render their analysis infeasible, especially when considering high-resolution data or analyses across multiple subjects or conditions. Here, we present an object-based matrix representation that achieves a very low memory footprint by computing matrix elements on demand instead of explicitly storing them. In doing so, memory required for a dense connectome is reduced to the amount needed to store the underlying time series data. Based on theoretical considerations and benchmarks, different matrix object implementations and additional programs (based on available Matlab functions and Matlab-based third-party software) are compared with regard to their computational efficiency. The matrix implementation based on on-demand computations has very low memory requirements, thus enabling analyses that would be otherwise infeasible to conduct due to insufficient memory. An open source software package containing the created programs is available for download.

More than meets the eye: Raw scanpath replay is an insufficient memory cue for static and dynamic scenes.

More than meets the eye: Raw scanpath replay is an insufficient memory cue for static and dynamic scenes.

Acquisition of earthworm-like movement patterns of many-segmented peristaltic crawling robots

In recent years, attention has been increasingly devoted to the development of rescue robots that can protect humans from the inherent risks of rescue work. Particularly, anticipated is the development of a robot that can move deeply through small spaces. We have devoted our attention to peristalsis, the movement mechanism used by earthworms. A reinforcement learning technique used for the derivation of the robot movement pattern, Q-learning, was used to develop a three-segmented peristaltic crawling robot with a motor drive. Characteristically, peristalsis can provide movement capability if at least three segments work, even if a segmented part does not function. Therefore, we had intended to derive the movement pattern of many-segmented peristaltic crawling robots using Q-learning. However, because of the necessary increase in calculations, in the case of many segments, Q-learning cannot be used because of insufficient memory. Therefore, we devoted our attention to a learning method called Actor–Critic, which can be implemented with low memory. Because Actor-Critic methods are TD methods that have a separate memory structure to explicitly represent the policy independent of the value function. Using it, we examined the movement patterns of six-segmented peristaltic crawling robots.