Limited Amount Of Memory Research Articles

Multi-scale adaptive weighted network for polarization computational imaging super-resolution

Due to the pixel limit of the polarization imaging detector and the object detection conditions, the spatial resolution of the object polarization image in actual imaging detection applications is generally low. Convolutional neural networks (CNNs) have been introduced to image super-resolution (SR). However, these methods rarely explore the internal connections between local spatial components and multi-scale feature maps in different receptive fields. To this end, we propose a multi-scale adaptive weighted network (MSAWN) for polarization computational imaging super-resolution to gain superior reconstruction performance. Computational imaging methods centered on information acquisition and interpretation can obtain high-resolution images that are superior to imaging systems. Specifically, we use a limited amount of memory and computational power even with multi-scale and multi-level polarization information. Second, a spatial pyramid structure based on the space-channel attention mechanism is designed to effectively adjust the feature weight of polarization information. Third, we adopt an adaptive weight unit to reduce redundant network branches and parameters. Particularly, we design an innovative reconstruction layer with inputs coming from multiple paths by means of sub-pixel convolution. The experimental results show that the proposed method achieves better reconstruction accuracy and visual effect, and the objective evaluation indexes such as PSNR and SSIM are significantly improved.

РЕАЛИЗАЦИЯ НЕЛИНЕЙНОГО ПРЕОБРАЗОВАНИЯ ПОДСТАНОВКИ В ВИДЕ ЭЛЕМЕНТОВ СИСТЕМЫ СЧИСЛЕНИЯ РЯДА ФАКТОРИАЛЬНЫХ МНОЖЕСТВ

The paper investigates a nonlinear substitution transformation, presented as an element of the number system of factorial sets. This approach takes into account the characteristic features of substitutions and al-lows you to minimize the amount of memory for storing. Based on the unambiguous numbering approach for substitutions, three methods of storing in memory are considered: in the form of a vector, in the form of values of cyclic shifts of elements of an identical substitution, and in the form of a decimal substitution number. The results of the analysis of the amount of data required for storing the nonlinear substitution transformation for cryptographic primitives are presented. The complexity of translating the transformed substitutions into the standard form of storage by a vector is determined. The obtained results can be ap-plied in software solutions for ensuring information security using cryptographic primitives containing a nonlinear substitution transformation. The proposed approach to writing in the form of substitution trans-formations is recommended for implementation in complexes with a limited amount of memory.

On the Feasibility of Fast Fourier Transform Separability Property for Distributed Image Processing

Given the high algorithmic complexity of applied-to-images Fast Fourier Transforms (FFT), computational-resource-usage efficiency has been a challenge in several engineering fields. Accelerator devices such as Graphics Processing Units are very attractive solutions that greatly improve processing times. However, when the number of images to be processed is large, having a limited amount of memory is a serious problem. This can be faced by using more accelerators or using higher-capability accelerators, which implies higher costs. The separability property is a resource in hardware approaches that is frequently used to divide the two-dimensional FFT work into several one-dimensional FFTs, which can be simultaneously processed by several computing units. Then, a feasible alternative to address this problem is distributed computing through an Apache Spark cluster. However, determining the separability-property feasibility in distributed systems, when migrating from hardware implementations, is not evident. For this reason, in this paper a comparative study is presented between distributed versions of two-dimensional FFTs using the separability property to determine the suitable way to process large image sets using both Spark RRDs and DataFrame APIs.

Optimal retrieval in puzzle‐based storage with heuristic search and tabulation

AbstractPuzzle‐based storage (PBS) systems provide a space efficient solution for storage and retrieval of unit loads (e.g., cars, boxes) in the very land‐expensive urban environment. In this work, we formulate a highly generic optimal retrieval problem, consisting of any number of loads to be retrieved, and an arbitrary number of I/O points and empty locations (escorts) on a two‐dimensional lattice graph. We theoretically analyze the characteristics of the problem and propose a set of graph search algorithms to tackle the inherent complexities thereof. In contrast to existing work in the field, we investigate the runtime‐memory tradeoff pertaining to our problem, highlighting the possible benefits and limitations of our heuristic search and tabulation techniques. Our results demonstrate that our algorithms are capable of optimally solving moderate‐sized PBS problems using a limited amount of memory and in a reasonable amount of time.

Tiered Sampling

We introduce Tiered Sampling , a novel technique for estimating the count of sparse motifs in massive graphs whose edges are observed in a stream. Our technique requires only a single pass on the data and uses a memory of fixed size M , which can be magnitudes smaller than the number of edges. Our methods address the challenging task of counting sparse motifs—sub-graph patterns—that have a low probability of appearing in a sample of M edges in the graph, which is the maximum amount of data available to the algorithms in each step. To obtain an unbiased and low variance estimate of the count, we partition the available memory into tiers (layers) of reservoir samples. While the base layer is a standard reservoir sample of edges, other layers are reservoir samples of sub-structures of the desired motif. By storing more frequent sub-structures of the motif, we increase the probability of detecting an occurrence of the sparse motif we are counting, thus decreasing the variance and error of the estimate. While we focus on the designing and analysis of algorithms for counting 4-cliques, we present a method which allows generalizing Tiered Sampling to obtain high-quality estimates for the number of occurrence of any sub-graph of interest, while reducing the analysis effort due to specific properties of the pattern of interest. We present a complete analytical analysis and extensive experimental evaluation of our proposed method using both synthetic and real-world data. Our results demonstrate the advantage of our method in obtaining high-quality approximations for the number of 4 and 5-cliques for large graphs using a very limited amount of memory, significantly outperforming the single edge sample approach for counting sparse motifs in large scale graphs.

A Comparison of Limited-memory Krylov Methods for Stieltjes Functions of Hermitian Matrices

Given a limited amount of memory and a target accuracy, we propose and compare several polynomial Krylov methods for the approximation of $f(A){b}$, the action of a Stieltjes matrix function of a l...

Implementation of Numerical Integration to High-Order Elements on the GPUs

This article presents ways to implement a resource-consuming algorithm on hardware with a limited amount of memory, which is the GPU. Numerical integration for higher-order finite element approximation was chosen as an example algorithm. To perform computational tests, we use a non-linear geometric element and solve the convection-diffusion-reaction problem. For calculations, a Tesla K20m graphics card based on Kepler architecture and Radeon r9 280X based on Tahiti XT architecture were used. The results of computational experiments were compared with the theoretical performance of both GPUs, which allowed an assessment of actual performance. Our research gives suggestions for choosing the optimal design of algorithms as well as the right hardware for such a resource-demanding task.

Empowering Density-based Micro-clusters In Dynamic Data Stream Clustering

Data stream is a continuous sequence of data generated from various sources and continuously transferred from source to target. Streaming data needs to be processed without having access to all of the data. Some of the sources generating data streams are social networks, geospatial services, weather monitoring, e-commerce purchases, etc. Data stream mining is the process of acquiring knowledge structures from the continuously arriving data. Clustering is an unsupervised machine learning technique that can be used to extract knowledge patterns from the data stream. The mining of streaming data is challenging because the data is in huge amounts and arriving continuously. So the traditional algorithms are not suitable for mining data streams. Data stream mining requires fast processing algorithms using a single scan and a limited amount of memory. The micro clustering has a good role in this. In itself, density based micro clustering has its own unique place in data stream mining. This paper presents a survey on different data clustering algorithms, realizes and empowers the use of density-based micro clusters.

Fast and memory-efficient algorithms for high-order Tucker decomposition

Multi-aspect data appear frequently in web-related applications. For example, product reviews are quadruplets of the form (user, product, keyword, timestamp), and search-engine logs are quadruplets of the form (user, keyword, location, timestamp). How can we analyze such web-scale multi-aspect data on an off-the-shelf workstation with a limited amount of memory? Tucker decomposition has been used widely for discovering patterns in such multi-aspect data, which are naturally expressed as large but sparse tensors. However, existing Tucker decomposition algorithms have limited scalability, failing to decompose large-scale high-order ($$\ge $$ 4) tensors, since they explicitly materialize intermediate data, whose size grows exponentially with the order. To address this problem, which we call “Materialization Bottleneck,” we propose S-HOT, a scalable algorithm for high-order Tucker decomposition. S-HOT minimizes materialized intermediate data by using an on-the-fly computation, and it is optimized for disk-resident tensors that are too large to fit in memory. We theoretically analyze the amount of memory and the number of data scans required by S-HOT. Moreover, we empirically show that S-HOT handles tensors with higher order, dimensionality, and rank than baselines. For example, S-HOT successfully decomposes a real-world tensor from the Microsoft Academic Graph on an off-the-shelf workstation, while all baselines fail. Especially, in terms of dimensionality, S-HOT decomposes 1000 $$\times $$ larger tensors than baselines.

CMSS: Sketching based reliable tracking of large network flows

Abstract Reliably tracking large network flows in order to determine so-called elephant flows, also known as heavy hitters or frequent items, is a common data mining task. Indeed, this kind of information is crucial in many different contexts and applications. Since storing all of the traffic is impossible, owing to the fact that flows arrive as an unbounded, infinite length stream, many different algorithms have been designed for traffic measurement using only a limited amount of memory. CountMax is a recently published algorithm that solves this problem approximately by combining ideas from the CountMin and MisraGries algorithms. In this paper we introduce CMSS which cleverly combines ideas from the CountMin and Space Saving algorithms. We compare CMSS and CountMax on both synthetic and real datasets. The experimental results show that both algorithms achieve 100% recall, and can retrieve the frequent items without false negatives even with a limited amount of budgeted memory. Regarding the precision, CMSS proves to be robust and independent from all of the parameters, whilst CountMax is severely affected by false positives. Finally, both algorithms provide, in practice, the same frequency estimation quality. It follows that CMSS outperforms CountMax with regard to overall accuracy whilst providing the same frequency estimation quality.

Streaming PCA and Subspace Tracking: The Missing Data Case.

For many modern applications in science and engineering, data are collected in a streaming fashion carrying time-varying information, and practitioners need to process them with a limited amount of memory and computational resources in a timely manner for decision making. This often is coupled with the missing data problem, such that only a small fraction of data attributes are observed. These complications impose significant, and unconventional, constraints on the problem of streaming Principal Component Analysis (PCA) and subspace tracking, which is an essential building block for many inference tasks in signal processing and machine learning. This survey article reviews a variety of classical and recent algorithms for solving this problem with low computational and memory complexities, particularly those applicable in the big data regime with missing data. We illustrate that streaming PCA and subspace tracking algorithms can be understood through algebraic and geometric perspectives, and they need to be adjusted carefully to handle missing data. Both asymptotic and non-asymptotic convergence guarantees are reviewed. Finally, we benchmark the performance of several competitive algorithms in the presence of missing data for both well-conditioned and ill-conditioned systems.

Optimal mapping of program overlays onto many-core platforms with limited memory capacity

This paper addresses the problem of mapping tasks onto an FPGA-based many-core platform where the cores typically have a limited amount of memory and thus should be frequently overlaid with a small program block that implements a task. In this regard, we propose a framework that takes integer linear programming (ILP) to find an optimal mapping of an application onto such a many-core platform at the task-level of granularity. The optimality is defined within the limits of our ILP model. The proposed framework is not only suitable for an application that can be accommodated on the available cores but also for a larger application (or even multiple applications) that needs more cores than what is provided by the platform. This is achieved by mapping different partitions of the application to the same set of cores and dynamically (during the life time of the application) overlaying a partition on another. The proposed mapping flow integrates scheduling, binding and place and route steps into one mapping process using an ILP formulation. Due to the slowness of ILP solutions, our solution is applicable at design time only. It is implemented using TOMLAB/CPLEX toolbox and we assess its efficacy on a set of 40 synthetic task graphs as well as some multimedia applications.

Recyclable Counter With Confinement for Real-Time Per-Flow Measurement

With the amount of Internet traffic increasing substantially, measuring per-flow traffic accurately is an important task. Because of the nature of high-speed routers, a measurement algorithm should be fast enough to process every packet going through them, and should be executable with only a limited amount of memory, as well. In this paper, we use two techniques to solve memory/speed constraints: (1) recycling a memory block by resetting it (for memory constraint), and (2) confinement of virtual vectors to one word (for speed constraint). These techniques allow our measurement algorithm, called a recyclable counter with confinement (RCC), to accurately measure all individual flow sizes with a small amount of memory. In terms of encoding speed, it uses about one memory access and one hash computation. Unlike other previously proposed schemes, RCC decodes very quickly, demanding about three memory accesses and two hash calculations. This fast decoding enables real-time detection of a high uploader/downloader. Finally, RCC's data structure includes flow labels for large flows, so it is possible to quickly retrieve a list of large-flow names and sizes.

Morphable Resistive Memory Optimization for Mobile Virtualization

Virtualization offers significant benefits, such as better isolation and security for mobile systems. However, the limited amount of memory and virtualization’s memory-demanding nature make it challenging to virtualize mobile systems efficiently. In this paper, we utilize morphable resistive memories to design a high-performance mobile system with an extensible memory space. With morphable resistive memories, a simple and effective page management technique, Balloonfish, is proposed to convert the memory cell state between multilevel and single-level for achieving a balance between performance and memory space. First, an application-specific page allocation is proposed for managing morphable resistive memories in virtualized mobile systems. Besides, we use a balloon-style algorithm to balance memory allocation among multiple virtual machines. Our evaluation based on the Samsung Exynos 5250 system-on-chip with various real Android applications shows that our system achieves 28.63% performance improvement compared with the baseline scheme.

Cluster Head Election with Hexagonal Node Deployment Technique in Wireless Sensor Networks

Wireless Sensor Networks (WSNs) are a network of large number of wireless sensor nodes deployed over a wide geographical area. The sensor nodes have a limited amount of memory as well as power. Excess extent of energy is dissipated during data transmission to the base station (sink) from normal sensor nodes. The efficient way of prolonging the lifetime of sensor nodes is clustering which optimizes the energy dissipation of WSNs. In this paper, a homogenous clustering algorithm based on fuzzy logic has been proposed in which cluster head election is done with the help of two fuzzy descriptors namely residual energy and proximity distance. The sensor nodes have been hexagonally deployed in the homogenous environment. This covers the sensing area efficiently along with reducing the redundancy in data being transmitted by the sensor nodes. This algorithm has been compared with the low energy adaptive clustering hierarchy (LEACH) protocol. The simulation results show that the proposed approach outperforms the LEACH protocol in terms of number of data packets transmitted to base station, energy consumption and prolonged network lifetime.

Triangle Counting in Dynamic Graph Streams

Estimating the number of triangles in graph streams using a limited amount of memory has become a popular topic in the last decade. Different variations of the problem have been studied, depending on whether the graph edges are provided in an arbitrary order or as incidence lists. However, with a few exceptions, the algorithms have considered insert-only streams. We present a new algorithm estimating the number of triangles in dynamic graph streams where edges can be both inserted and deleted. We show that our algorithm achieves better time and space complexity than previous solutions for various graph classes, for example sparse graphs with a relatively small number of triangles. Also, for graphs with constant transitivity coefficient, a common situation in real graphs, this is the first algorithm achieving constant processing time per edge. The result is achieved by a novel approach combining sampling of vertex triples and sparsification of the input graph. In the course of the analysis of the algorithm we present a lower bound on the number of pairwise independent 2-paths in general graphs which might be of independent interest. At the end of the paper we discuss lower bounds on the space complexity of triangle counting algorithms that make no assumptions on the structure of the graph.

Estimating life-time distribution by observing population continuously

Often in a real world system with a fairly large population, members are not individually traceable for various reasons. As a result, the relationship between a member’s behavior and the system’s behavior is quite hard to understand. In this paper, we focus on a fundamental problem in such a system: the relationship between its population size and the life-time distribution for its members. We answer two questions: (A)If the life-time distribution is known and the times when members join are observable, how do we best estimate the population size?(B)If the population size can be observed accurately, how do we estimate the unknown life-time distribution for members? In the paper we focus on (B), using the results of (A) as a basis.We model the system as a G/GI/∞ queue with incomplete information, where jobs, once entering the queue, are no longer tracked. With this model, the population size is the number of jobs in the queue and the life times of members are the service times of the jobs. The problem (A) is to estimate the number of jobs in the queue, with known arrival times and a known service-time distribution. We show that, in terms of mean square error, the best deterministic estimator for the (stochastic) number of jobs in the system can be constructed using the survival function of the service-time distribution. The problem (B) is to estimate the unknown service-time distribution in a G/GI/∞ queue where the number of jobs are observable. We demonstrate that the service-time distribution can be inferred indirectly from continuous observations of the number of jobs in the queue, and then propose a few easy-to-implement algorithms. Using only a limited amount of memory, these on-line, streaming algorithms continuously refine their results which, over time, converge to the true service-time distribution.

A repository for scalable model management

Applying model-driven engineering (MDE) in industrial-scale systems requires managing complex models which may be very large. These models must be persisted in a way that allows their manipulation by client applications without fully loading them. In this paper, we propose Morsa, a model repository that provides scalable manipulation of large models through load on demand and incremental store; model persistence is supported by a NoSQL database. We discuss some load on demand and incremental store algorithms as well as a database design. A prototype that integrates transparently with EMF is presented, and its evaluation demonstrates that it is capable of fully managing large models with a limited amount of memory. Moreover, a set of benchmarks has been executed, exhibiting better performance than the EMF XMI file-based persistence and the most widely used model repository, CDO.

Rapid change in environment for research on diagnostic ultrasound.

For progression of research, of course, experimental systems play an important role. Fifteen years ago, at the start of my research career, it was still difficult for researchers, maybe particularly for researchers in universities, to obtain an experimental system that was capable of acquiring beamformed ultrasonic radio-frequency (RF) echoes from an ultrasonic array transducer, due to the limitations of the available hardware, e.g., low sampling frequency and the limited memory of the analog-to-digital (A/D) converter. In our case, we were forced to use base-band signals (composed of in-phase and quadrature signals) to acquire ultrasonic signals for longer acquisition periods with a limited amount of memory because acquisition of beamformed RF signals required a higher sampling frequency and more memory. Even for acquisition of base-band signals, we needed to develop an expensive in-house acquisition system to acquire the base-band signals of beamformed ultrasonic RF echoes for several seconds. The sampling frequency and amount of memory for the in-house system were 10 MHz and 16 MB, respectively. There was a severe tradeoff between sampling frequency and the period available for acquisition (amount of memory). For example, for research on ultrasound imaging, it is desirable to acquire RF signals, which requires a high sampling frequency, and, thus, the number of frames acquired is reduced. On the other hand, in the case of research involving measurement of tissue motion, for example, it is necessary to acquire ultrasonic signals for a certain required period and the sampling frequency should be limited. Therefore, signals such as base-band signals were preferred because they did not require a high sampling frequency. In the past decade, the capability of acquisition systems has improved dramatically. Now we are able to obtain an acquisition system with a much higher sampling frequency and more memory. At around the year 2000, such systems implemented acquisition of beamformed RF signals for a certain required number of frames, e.g., for several seconds. However, it was still difficult to acquire ultrasonic RF echoes received by each element of an ultrasonic array transducer (channel RF signal). Therefore, there were very few research groups who could develop and implement methods for diagnostic ultrasound imaging using RF signals obtained from array elements, though there should have been many good ideas on diagnostic ultrasound imaging based on array signal processing. However, systems capable of acquiring channel RF signals and also controlling the transmitted ultrasonic field have become available at last due to further innovation in the capability of acquisition systems. Such systems are still not cheap, but now we are able to obtain such systems much more easily than before. As described above, there was a severe tradeoff in the past, i.e., higher sampling frequency and shorter acquisition period to acquire RF echoes or lower sampling frequency and longer acquisition period based on base-band signals. However, now we have been freed from such limitations. Current ultrasound acquisition systems, which are able to acquire channel RF signals and control the transmitted ultrasonic field, will significantly broaden the opportunities for research on diagnostic ultrasound imaging, and the prevalence of such systems may lead to explosive progression in related research fields. H. Hasegawa (&) Graduate School of Biomedical Engineering, Tohoku University, Sendai, Miyagi, Japan e-mail: hasegawa@ecei.tohoku.ac.jp

A sliding window based algorithm for frequent closed itemset mining over data streams

Frequent pattern mining over data streams is an important problem in the context of data mining and knowledge discovery. Mining frequent closed itemsets within sliding window instead of complete set of frequent itemset is very interesting since it needs a limited amount of memory and processing power. Moreover, handling concept change within a compact set of closed patterns is faster. However, it requires flexible and efficient data structures as well as intuitive algorithms. In this paper, we have introduced an effective and efficient algorithm for closed frequent itemset mining over data streams operating in the sliding window model. This algorithm uses a novel data structure for storing transactions of the window and corresponding frequent closed itemsets. Moreover, the support of a new frequent closed itemset is efficiently computed and an old pattern is removed from the monitoring set when it is no longer frequent closed itemset. Extensive experiments on both real and synthetic data streams show that the proposed algorithm is superior to previously devised algorithms in terms of runtime and memory usage.