Search

Communication overhead is one of the major performance bottlenecks in large-scale distributed computing systems, in particular for machine learning applications. Conventionally, compression techniques are used to reduce the load of communication by combining intermediate results of the same computation task as much as possible. Recently, via the development of coded distributed computing (CDC), it has been shown that it is possible to enable coding opportunities across intermediate results of different computation tasks to further reduce the communication load. We propose a new scheme, named compressed coded distributed computing (in short, compressed CDC ), which jointly exploits the above two techniques (i.e., combining the intermediate results of the same computation and coding across the intermediate results of different computations) to significantly reduce the communication load for computations with linear aggregation (reduction) of intermediate results in the final stage that are prevalent in machine learning (e.g., distributed training algorithms where partial gradients are computed distributedly and then averaged in the final stage). In particular, compressed CDC first compresses/combines several intermediate results for a single computation, and then utilizes multiple such combined packets to create a coded multicast packet that is simultaneously useful for multiple computations. We characterize the achievable communication load of compressed CDC and show that it substantially outperforms both combining methods and CDC scheme. Based on the compressed CDC technique, we then study a distributed training problem as one of its application. We characterize the communication load for this distributed training problem and show that it is asymptotically optimal.

We study two problems of private matrix multiplication, over a distributed computing system consisting of a master node, and multiple servers that collectively store a family of public matrices using Maximum-Distance-Separable (MDS) codes. In the first problem of Private and Secure Matrix Multiplication (PSMM) from colluding servers, the master intends to compute the product of its confidential matrix $\mathbf{A}$ with a target matrix stored on the servers, without revealing any information about $\mathbf{A}$ and the index of target matrix to some colluding servers. In the second problem of Fully Private Matrix Multiplication (FPMM) from colluding servers, the matrix $\mathbf{A}$ is also selected from another family of public matrices stored at the servers in MDS form. In this case, the indices of the two target matrices should both be kept private from colluding servers. We develop novel strategies for the two PSMM and FPMM problems, which simultaneously guarantee information-theoretic data/index privacy and computation correctness. We compare the proposed PSMM strategy with a previous PSMM strategy with a weaker privacy guarantee (non-colluding servers), and demonstrate substantial improvements over the previous strategy in terms of communication and computation overheads. Moreover, compared with a baseline FPMM strategy that uses the idea of Private Information Retrieval (PIR) to directly retrieve the desired matrix multiplication, the proposed FPMM strategy significantly reduces storage overhead, but slightly incurs large communication and computation overheads., 16 pages, 6 figures

Individuals with autism spectrum disorder (ASD) experience spatial working memory deficits and show different encoding mechanisms from typical developing (TD) peers. To effectively describe the encoding strategies of those with ASD and highlight their characteristics in cognitive processing, we adopted improved change detection tasks and added eye-movement indicators to investigate the chunking function and reference preference of children with and without ASD. The current study included 20 participants with ASD aged 8-16 and 20 TD children matched for age, gender, and intelligence. Experiment 1 used high/low-structured change detection tasks, and eye-movement indexes were recorded as they memorized the locations of the items to investigate spatial chunking strategies. In Experiment 2, changes in eye movement patterns were observed by adding a frame of reference. The results suggested different encoding strategies in ASD and TD individuals. The ASD group showed local processing bias and had difficulty adopting chunking strategies in spatial working memory. Eye-movement analysis suggested that they rarely showed integrated information processing tendency observed in TD children. Moreover, as a compensatory processing, they were more likely to use the frame of reference. In this study, we compared the spatial chunking strategies and reference preference of children with and without ASD, and eye-movement analysis was used to investigate the processing mechanism. These findings are significant for research on cognitive characteristics of ASD and provide a new focus for working memory training in children with ASD. LAY SUMMARY: The current study suggests that children with autism spectrum disorder are poorer at organizing items into chunks in spatial working memory, but rely more on reference frames. If the purpose of location memory is to strengthen the adaptability of children with autism, it should provide them with more clues or references. If it is for the purpose of intervention such as cognitive training, it should guide them to integrate information to improve the basic cognitive processing efficiency. Autism Res 2021, 14: 897-910. © 2020 International Society for Autism Research and Wiley Periodicals LLC.

Understanding the oil emplacement influence on diagenesis is one of the most key factors of reservoir quality prediction. The Devonian Donghe sandstones are deeply buried to 5700～6000 m, and they are important exploration target in the Tarim Basin, additional the reservoirs experienced multi-phase hydrocarbon accumulation and destruction. The Donghe sandstones contain various diagenetic minerals, which are mainly composed of calcite cement, a small amount of quartz overgrowth and pyrite, trace ankerite and ferrocalctie. The current study investigates the influence of multiple-stage oil emplacement on the diagenesis with particular attention to calcite and quartz cementation of marine clastic reservoir interval of the Donghe sandstones in the Tarim Basin. The samples in the oil leg and water leg have been collected and a series of studies were performed, including thin section petrography, scanning electron microscopy (SEM), carbon and oxygen isotopes and cathodoluminescence (CL) analysis. The δ13CV-PDB values of calcite cement from the oil leg and water leg between −4.73 and −1.21‰ and −3.81 to −0.86‰, respectively. The δ18OV-PDB values of calcite cement from the oil leg and water leg range from −15.59 to −6.61‰ and −14.97 to −7.76‰, respectively. Both of them display no obviously different in the oil leg and water leg. All of the δ13CV-PDB values in the whole interval of the Donghe sandstones show lightly depletion was explained mixing from marine carbon and carbon produced from bacterial sulfate reduction (BSR) that happens during petroleum degradation associated with sulfate reduced by hydrocarbons bacterially. Two dull-luminescence zones, trace Fe and Mn content, and high Fe accompanied by low Mn content, and orange-red- and bright-orange-yellow-luminescence zones in oil leg indicate four generations of calcite occurred, whereas two generations of calcite occurs in water leg, which are dull- and orange-red-luminescence zones. Two origins of calcite cement were clearly identified: marine biogenic carbonates and hydrocarbon-related from palaeo-reservoir BSR during uplift. The first generation of calcite cement originated from biogenic carbonate, the second and third generations of calcite cement were mixing from predate carbonate cement dissolved and re-precipitated, and calcite precipitated by BSR directly. Some quartz grains and K-feldspars were dissolved by residual organic acids that remained in the irreducible water, and following replaced by the fourth generation of calcite cement. Some calcite cement in the oil leg precipitated in a temperature higher than the homogeneous temperature of fluid inclusion, which indicates oil emplacement has no influence on calcite cementation. The quartz overgrowth postdate the first hydrocarbon accumulation and predate the second hydrocarbon accumulation, which originated from quartz grains dissolved by organic acids that produced from BSR and re-precipitated at a low temperature ( The current study can promote the understanding of the influence of multiple-stage hydrocarbon accumulation on diagenesis, and it can also inspire the diagenesis process of other Paleozoic hydrocarbon reservoirs in the Tarim Basin.

—The Carboniferous Donghe sandstone reservoir is the most important target in the Tabei Uplift of the Tarim Basin, which contains a range of hydrocarbon types, including bitumen, heavy oil, condensate oil, light oil, crude oil, and hydrocarbon gas, and has high contents of CO2 and N2. The origin of multiple phase hydrocarbons from Carboniferous reservoir rocks in the Donghetang area, Western Tabei Uplift, is documented in this paper based on integral analysis of the geochemistry, pyrolysis, and carbon isotopes of the bulk composition and light composition hydrocarbons. Oil–source correlations determined that the paleoreservoir hydrocarbons that formed from Permian to Triassic derived from the Lower Ordovician (O1) source rocks and that those of the present-day reservoir that formed in the Neogene derived from Middle–Upper Ordovician (O2-3) source rocks. During the uplift episode lasting from Permian to Triassic, the hydrocarbons in the entire paleoreservoir underwent water washing, biodegradation, and bacterial sulfate reduction (BSR), resulting in residual bitumen, heavy oil, H2S, and pyrites in the paleoreservoir. The high CO2 and N2 contents originated from volcanic degassing due to volcanic activity from Permian to Early Triassic. The present-day reservoirs underwent gas washing and evaporative fractionation due to natural gas charging that originated from oil cracking and kerogen degradation in the deeper reservoirs; this resulted in fractionation and formed condensate oil and light oil with a high wax content in the residual crude oil. Based on this research, it was concluded that the diverse hydrocarbon phases in the Donghetang area were primarily attributed to water washing, biodegradation, BSR, volcanic degassing, gas washing, and evaporative fractionation.

Psychological studies have generally shown that individuals with Autism Spectrum Disorder (ASD) have specificity in the processing of social information by using static or abstract images. Yet, a recent study showed that there was no differ in their use of the added social or non-social cues in dynamic interactive situations. To establish the cause of the inconsistent results, we explored the processing pattern of gaze cues in individuals with ASD by using chase detection paradigm and eye-tracking methodologies. In this study, unlike typical controls, participants with ASD showed no detection advantage under the oriented condition. The results suggested that individuals with ASD may utilize an atypical processing pattern in dynamic interactive situations.

The problem of $X$-secure $T$-colluding symmetric Private Polynomial Computation (PPC) from coded storage system with $B$ Byzantine and $U$ unresponsive servers is studied in this paper. Specifically, a dataset consisting of $M$ files is stored across $N$ distributed servers according to $(N,K+X)$ Maximum Distance Separable (MDS) codes such that any group of up to $X$ colluding servers can not learn anything about the data files. A user wishes to privately evaluate one out of a set of candidate polynomial functions over the $M$ files from the system, while guaranteeing that any $T$ colluding servers can not learn anything about the identity of the desired function and the user can not learn anything about the $M$ data files more than the desired polynomial function evaluations, in the presence of $B$ Byzantine servers that can send arbitrary responses maliciously to confuse the user and $U$ unresponsive servers that will not respond any information at all. A novel symmetric PPC scheme using Lagrange encoding is proposed. This scheme achieves a PPC rate of $1-\frac{G(K+X-1)+T+2B}{N-U}$ with secrecy rate $\frac{G(K+X-1)+T}{N-(G(K+X-1)+T+2B+U)}$ and finite field size $N+\max\{K,N-(G(K+X-1)+T+2B+U)\}$, where $G$ is the maximum degree over all the candidate polynomial functions. Moreover, to further measure the efficiency of PPC schemes, upload cost, query complexity, server computation complexity and decoding complexity required to implement the scheme are analyzed. Remarkably, the PPC setup studied in this paper generalizes all the previous MDS coded PPC setups and the degraded schemes strictly outperform the best known schemes in terms of (asymptotical) PPC rate, which is the main concern of the PPC schemes.

Silicon nitride (Si3N4) with a high α-phase content is in demand due to its higher thermal conductivity for use in heat dissipation bases for electronic devices. However, more needs to be understood about its fibrous growth, therefore, rich fibrous morphologies were synthesized with the assistance of Fe powder and NH4Cl. The effects of these additives on the phase content and fiber morphologies of the products were investigated The results illustrate that the Si3N4 products possess upper and lower layers and that when the Fe powder content is 4 wt% and NH4Cl content is 5 wt%, the Si3N4 fibers have smooth surfaces, uniform diameters, and no floating particles. Moreover, the maximum α-Si3N4 content reached 94.8 wt%, demonstrating an increase in this phase. The double mechanisms of vapor-liquid-solid and vapor-solid are presented as the growth mechanism at the fiber.

How can we optimally trade extra computing power to reduce the communication load in distributed computing? We answer this question by characterizing a fundamental tradeoff between computation and communication in distributed computing, i.e., the two are inversely proportional to each other. More specifically, a general distributed computing framework, motivated by commonly used structures like MapReduce, is considered, where the overall computation is decomposed into computing a set of "Map" and "Reduce" functions distributedly across multiple computing nodes. A coded scheme, named "Coded Distributed Computing" (CDC), is proposed to demonstrate that increasing the computation load of the Map functions by a factor of $r$ (i.e., evaluating each function at $r$ carefully chosen nodes) can create novel coding opportunities that reduce the communication load by the same factor. An information-theoretic lower bound on the communication load is also provided, which matches the communication load achieved by the CDC scheme. As a result, the optimal computation-communication tradeoff in distributed computing is exactly characterized. Finally, the coding techniques of CDC is applied to the Hadoop TeraSort benchmark to develop a novel CodedTeraSort algorithm, which is empirically demonstrated to speed up the overall job execution by $1.97\times$ - $3.39\times$, for typical settings of interest., To appear in IEEE Transactions on Information Theory

In this paper, we propose coded Merkle tree (CMT), a novel hash accumulator that offers a constant-cost protection against data availability attacks in blockchains, even if the majority of the network nodes are malicious. A CMT is constructed using a family of sparse erasure codes on each layer, and is recovered by iteratively applying a peeling-decoding technique that enables a compact proof for data availability attack on any layer. Our algorithm enables any node to verify the full availability of any data block generated by the system by just downloading a \(\varTheta (1)\) byte block hash commitment and randomly sampling \(\varTheta (\log b)\) bytes, where b is the size of the data block. With the help of only one connected honest node in the system, our method also allows any node to verify any tampering of the coded Merkle tree by just downloading \(\varTheta (\log b)\) bytes. We provide a modular library for CMT in Rust and Python and demonstrate its efficacy inside the Parity Bitcoin client.

Aims Melanoma is a fatal malignancy. Karyopherin α 2 (KPNA2) plays an important role in many carcinogenesis. This study was aimed to study the role of KPNA2 in cellular functions and molecular mechanisms of melanoma. Main methods We investigated the expression and prognosis of KPNA2 in melanoma using the GEPIA database ( http://gepia.cancer-pku.cn/ ). The effect of KPNA2 on melanoma cells was determined using real-time PCR, western blot, immunofluorescence assay, CCK-8, colony formation, wound healing assay, transwell assay, EMSA, and immunohistochemistry. The influence of KPNA2 on the tumorigenicity of melanoma cells was evaluated in a nude mice model in vivo. Key findings Our results showed that KPNA2 expression is relatively high in melanoma tissues and cells, and melanoma patients with higher expression of KPNA2 had lower overall survival rate and disease free survival rate. KPNA2 promoted proliferation ability and increased the expression of PCNA, Ki67, and C-MYC in melanoma cells. Further, KPNA2 could promote migration and invasion and increase the expression of MMP2 and MMP9. Mechanism studies showed that KPNA2 activated NF-κB/p65 signaling pathways, as evidenced by the nuclear translocation of p65 and increased the expression of COX-2, ICAM-1, iNOS, and MCP1 in melanoma cells. NF-κB inhibitor JSH-23 could reverse the pro-tumor effects of KPNA2 on melanoma cells. Moreover, upregulation of KPNA2 facilitated the tumorigenicity of melanoma cells. Significance KPNA2 promotes proliferation, migration and invasion through enhancing NF-κB/p65 signaling pathways in melanoma cells. Our study suggests KPNA2 as a potential therapeutic target for the treatment of melanoma.

We introduce an information-theoretic framework, named Coded State Machine (CSM), to securely and efficiently execute multiple state machines on untrusted network nodes, some of which are Byzantine. The standard method of solving this problem is using State Machine Replication, which achieves high security at the cost of low efficiency. We propose CSM, which achieves the optimal linear scaling in storage efficiency, throughput, and security simultaneously with the size of the network. The storage efficiency is scaled via the design of Lagrange coded states and coded input commands that require the same storage size as their origins. The computational efficiency is scaled using a novel delegation algorithm, called INTERMIX, which is an information-theoretically verifiable matrix-vector multiplication algorithm of independent interest. Using INTERMIX, the network nodes securely delegate their coding operations to a single worker node, and a small group of randomly selected auditor nodes verify its correctness, so that computational efficiency can scale almost linearly with the network size, without compromising on security.

This study explored student abilities in applying conceptual knowledge when presented with structured performance tasks. Specifically, the study gauged proficiency in higher-order applications of students enrolled in earth and environmental science or biology. The student sample was drawn from a Redesigned STEM high school model where a tested performance assessment protocol was employed for the purposes of the investigation. It was determined that performance-based proficiency was not uniform within tasks and applications, but could be recognized through student artifacts of learning on a situational basis. Based on the findings of the study, several implications are highlighted.

We consider energy-efficient indoor lighting control for adapting to daylight and user presence changes. Light sensors located at ceiling luminaires and at workspace desks are used to provide illumination information feedback, along with user presence information from occupancy sensors at ceiling luminaires, to a lighting controller. The light sensors at the workspace transmit information at a lower update rate than the sensors at the ceiling. Using this sensing information, we propose a control method to adapt the dimming levels of the luminaires such that desired illumination levels are achieved at the workspace level. The performance of the proposed method is evaluated in an office lighting testbed.

Human gingival fibroblasts (HGFs) are responsible for connective tissue repair and scarring, and are exposed to mechanical forces under physiological and pathological conditions. The exact mechanisms underlying gingival tissue reconstruction under mechanical forces remain unclear. The present study aimfed to investigate the effects of mechanical forces on the proliferation and extracellular matrix synthesis in HGFs by establishing a 3‑dimensional (3D) HGF culture model using poly(lactide‑co‑glycolide) (PLGA) scaffolds. HGFs were cultured in 3D PLGA scaffolds and a mechanical force of 0, 5, 15, 25 or 35 g/cm2 was applied to HGFs for 24 h. A mechanical force of 25 g/cm2 induced the highest proliferation rate, and thus was selected for subsequent experiments. Cell viability was determined using the MTT assay at 0, 24, 48 and 72 h. The expression levels of type I collagen (COL‑1) and matrix metallopeptidase (MMP)‑1 were examined by reverse transcription‑quantitative polymerase chain reaction and ELISA, and transforming growth factor (TGF)‑β expression was evaluated by ELISA. The application of mechanical force on HGFs cultured on the 3D PLGA scaffolds resulted in a significant increase in cell proliferation and COL‑1 expression, as well as a decrease in MMP‑1 expression. A TGF‑β1 inhibitor was also applied, which attenuated the effects of mechanical force on HGF proliferation, and COL‑1 and MMP‑1 expression, thus suggesting that TGF‑β signaling pathways may mediate the mechanical force‑induced alterations observed in HGFs. In conclusion, these findings helped to clarify the mechanisms underlying mechanical force‑induced HGF proliferation and ECM synthesis, which may promote the development of targeted therapeutics to treat various diseases, including gingival atrophy caused by orthodontic treatment.

Communication overhead is one of the major performance bottlenecks in large-scale distributed computing systems, in particular for machine learning applications. Conventionally, compression techniques are used to reduce the load of communication by combining intermediate results of the same computation task as much as possible. Recently, via the development of coded distributed computing (CDC), it has been shown that it is possible to enable coding opportunities across intermediate results of different computation tasks to further reduce the communication load. We propose a new scheme, named compressed coded distributed computing (in short, compressed CDC), which jointly exploits the above two techniques (i.e., combining the intermediate results of the same computation and coding across the intermediate results of different computations) to significantly reduce the communication load for computations with linear aggregation (reduction) of intermediate results in the final stage that are prevalent in machine learning (e.g., distributed training algorithms where partial gradients are computed distributedly and then averaged in the final stage). In particular, compressed CDC first compresses/combines several intermediate results for a single computation, and then utilizes multiple such combined packets to create a coded multicast packet that is simultaneously useful for multiple computations. We characterize the achievable communication load of compressed CDC and show that it substantially outperforms both combining methods and CDC scheme., Comment: A shorter version to appear in ISIT 2018

Today’s blockchain designs suffer from a trilemma claiming that no blockchain system can simultaneously achieve decentralization, security, and performance scalability. For current blockchain systems, as more nodes join the network, the efficiency of the system (computation, communication, and storage) stays constant at best. A leading idea for enabling blockchains to scale efficiency is the notion of sharding: different subsets of nodes handle different portions of the blockchain, thereby reducing the load for each individual node. However, existing sharding proposals achieve efficiency scaling by compromising on trust - corrupting the nodes in a given shard will lead to the permanent loss of the corresponding portion of data. In this paper, we settle the trilemma by demonstrating a new protocol for coded storage and computation in blockchains. In particular, we propose PolyShard : “polynomially coded sharding” scheme that achieves information-theoretic upper bounds on the efficiency of the storage, system throughput, as well as on trust, thus enabling a truly scalable system. We provide simulation results that numerically demonstrate the performance improvement over state of the arts, and the scalability of the PolyShard system. Finally, we discuss potential enhancements, and highlight practical considerations in building such a system.

Student interest is associated with persistence in STEM courses of study (Maltese, Melki, & Wiebke, 2014). If peers decide, of their own accord, to discuss knowledge among each other outside of the classroom context, the behavior is indicative of deepening interest in the information being shared (Renninger & Hidi, 2002). Understanding outside classroom knowledge sharing behaviors among peers involved in a STEM course may help educators construct learning contexts that promote interest and persistence in STEM subjects. To that end, this study examined two important research questions: (1) what are the key factors that influence peer to peer knowledge sharing outside the classroom? and (2) what are the methods the student use to share content knowledge? In order to explore these questions, a qualitative study was designed to explore knowledge sharing between peers outside the classroom. A semi-structured interview protocol with eight students from a Mid-Atlantic community college was conducted to explore students’ perceptions of knowledge sharing between peers. Data were coded and analyzed by a group of researchers and themes were identified and theoretical and practical implications of the study were recorded. Several key facilitators of knowledge sharing were identified: self-efficacy, interpersonal relationships, interpersonal similarity and media richness. Implications for teachers are presented. Limitations and future research are included in the end of the study. Keywords: knowledge sharing, peer to peer, peer learning, knowledge transfer, content knowledge, college science teaching, community college

We consider a mobile edge computing scenario, in which mobile users (e.g., smartphones) offload their computation requests to the computing nodes distributed at the network edge (e.g., base stations). The edge nodes process the offloaded requests in the “computation phase”, and return the results to the users in the “communication phase” through wireless links. With emphasis on the role of coding in both computation and communication phases, we introduce new architectures for mobile edge computing to exploit the best of both and achieve three salient features: 1) Optimality: We show that coded computations can induce collaboration among the nodes in communication phase, and thus achieves minimum possible load of communication, without requiring extra computation, therefore achieves the minimum possible load of computation simultaneously. 2) Universality: The code design in the coded computation does not require channel state information in communication phase, and therefore it is universality optimum for any channel coefficient. 3) Mobility tolerance: The computation results can be delivered from any subset of the edge nodes with slightly more computations, thus mobility of the users can be dealt with effectively.

We consider a mobile edge computing problem, in which mobile users offload their computation tasks to computing nodes (e.g., base stations) at the network edge. The edge nodes compute the requested functions and communicate the computed results to the users via wireless links. For this problem, we propose a Universal Coded Edge Computing (UCEC) scheme for linear functions to simultaneously minimize the load of computation at the edge nodes, and maximize the physical-layer communication efficiency towards the mobile users. In the proposed UCEC scheme, edge nodes create coded inputs of the users, from which they compute coded output results. Then, the edge nodes utilize the computed coded results to create communication messages that zero-force all the interference signals over the air at each user. Specifically, the proposed scheme is universal since the coded computations performed at the edge nodes are oblivious of the channel states during the communication process from the edge nodes to the users., To Appear in ISIT 2017

Redundancy is abundant in Fog networks (i.e., many computing and storage points) and grows linearly with network size. We demonstrate the transformational role of coding in Fog computing for leveraging such redundancy to substantially reduce the bandwidth consumption and latency of computing. In particular, we discuss two recently proposed coding concepts, namely Minimum Bandwidth Codes and Minimum Latency Codes, and illustrate their impacts in Fog computing. We also review a unified coding framework that includes the above two coding techniques as special cases, and enables a tradeoff between computation latency and communication load to optimize system performance. At the end, we will discuss several open problems and future research directions., To appear in IEEE Communications Magazine, Issue on Fog Computing and Networking

Today's data centers have an abundance of computing resources, hosting server clusters consisting of as many as tens or hundreds of thousands of machines. To execute a complex computing task over a data center, it is natural to distribute computations across many nodes to take advantage of parallel processing. However, as we allocate more and more computing resources to a computation task and further distribute the computations, large amounts of (partially) computed data must be moved between consecutive stages of computation tasks among the nodes, hence the communication load can become the bottleneck. In this paper, we study the optimal allocation of computing resources in distributed computing, in order to minimize the total execution time in distributed computing accounting for both the duration of computation and communication phases. In particular, we consider a general MapReduce-type distributed computing framework, in which the computation is decomposed into three stages: \emph{Map}, \emph{Shuffle}, and \emph{Reduce}. We focus on a recently proposed \emph{Coded Distributed Computing} approach for MapReduce and study the optimal allocation of computing resources in this framework. For all values of problem parameters, we characterize the optimal number of servers that should be used for distributed processing, provide the optimal placements of the Map and Reduce tasks, and propose an optimal coded data shuffling scheme, in order to minimize the total execution time. To prove the optimality of the proposed scheme, we first derive a matching information-theoretic converse on the execution time, then we prove that among all possible resource allocation schemes that achieve the minimum execution time, our proposed scheme uses the exactly minimum possible number of servers.

Modern learning algorithms use gradient descent updates to train inferential models that best explain data. Scaling these approaches to massive data sizes requires proper distributed gradient descent schemes where distributed worker nodes compute partial gradients based on their partial and local data sets, and send the results to a master node where all the computations are aggregated into a full gradient and the learning model is updated. However, a major performance bottleneck that arises is that some of the worker nodes may run slow. These nodes a.k.a. stragglers can significantly slow down computation as the slowest node may dictate the overall computational time. We propose a distributed computing scheme, called Batched Coupon's Collector (BCC) to alleviate the effect of stragglers in gradient methods. We prove that our BCC scheme is robust to a near optimal number of random stragglers. We also empirically demonstrate that our proposed BCC scheme reduces the run-time by up to 85.4% over Amazon EC2 clusters when compared with other straggler mitigation strategies. We also generalize the proposed BCC scheme to minimize the completion time when implementing gradient descent-based algorithms over heterogeneous worker nodes.

We propose a unified coded framework for distributed computing with straggling servers, by introducing a tradeoff between "latency of computation" and "load of communication" for some linear computation tasks. We show that the coded scheme of [1]-[3] that repeats the intermediate computations to create coded multicasting opportunities to reduce communication load, and the coded scheme of [4], [5] that generates redundant intermediate computations to combat against straggling servers can be viewed as special instances of the proposed framework, by considering two extremes of this tradeoff: minimizing either the load of communication or the latency of computation individually. Furthermore, the latency-load tradeoff achieved by the proposed coded framework allows to systematically operate at any point on that tradeoff to perform distributed computing tasks. We also prove an information-theoretic lower bound on the latency-load tradeoff, which is shown to be within a constant multiplicative gap from the achieved tradeoff at the two end points., a shorter version to appear in NetCod 2016

In this paper, we first review the Coded Distributed Computing (CDC) framework, recently proposed to significantly slash the data shuffling load of distributed computing via coding, and then discuss the extension of the CDC techniques to cope with two major challenges in general distributed computing problems, namely the straggling servers and multistage computations. When faced with straggling servers in a distributed computing cluster, we describe a unified coding scheme that superimposes CDC with the Maximum-Distance-Separable (MDS) coding on computation tasks, which allows a flexible tradeoff between computation latency and communication load. On the other hand, for a general multistage computation task expressed as a directed acyclic graph (DAG), we propose a coded framework that given the load of computation on each vertex of the DAG, applies the generalized CDC scheme individually on each vertex to minimize the communication load.

We introduce a general distributed computing framework, motivated by commonly used structures like MapReduce, and formulate an information-theoretic tradeoff between computation and communication in such a framework. We characterize the optimal tradeoff to within a constant factor, for all system parameters. In particular, we propose a coded scheme, namely “Coded MapReduce” (CMR), which creates and exploits coding opportunities in data shuffling for distributed computing, reducing the communication load by a factor that is linearly proportional to the computation load. We then prove a lower bound on the minimum communication load, and demonstrate that CMR achieves this lower bound to within a constant factor. This result reveals a fundamental connection between computation and communication in distributed computing - the two are inverse-linearly proportional to each other.

To explore the effects of different concentrations of putrescine on proliferation, migration, and apoptosis of human umbilical vein endothelial cells (HUVECs).HUVECs were routinely cultured in vitro. The 3rd to the 5th passage of HUVECs were used in the following experiments. (1) Cells were divided into 500, 1 000, and 5 000 µg/mL putrescine groups according to the random number table (the same grouping method was used for following grouping), with 3 wells in each group, which were respectively cultured with complete culture solution containing putrescine in the corresponding concentration for 24 h. Morphology of cells was observed by inverted optical microscope. (2) Cells were divided into 0.5, 1.0, 5.0, 10.0, 50.0, 100.0, 500.0, 1 000.0 µg/mL putrescine groups, and control group, with 4 wells in each group. Cells in the putrescine groups were respectively cultured with complete culture solution containing putrescine in the corresponding concentration for 24 h, and cells in control group were cultured with complete culture solution with no additional putrescine for 24 h. Cell proliferation activity (denoted as absorption value) was measured by colorimetry. (3) Cells were divided (with one well in each group) and cultured as in experiment (2), and the migration ability was detected by transwell migration assay. (4) Cells were divided (with one flask in each group) and cultured as in experiment (2), and the cell apoptosis rate was determined by flow cytometer. Data were processed with one-way analysis of variance, Kruskal-Wallis test, and Dunnett test.(1) After 24-h culture, cell attachment was good in 500 µg/mL putrescine group, and no obvious change in the shape was observed; cell attachment was less in 1 000 µg/mL putrescine group and the cells were small and rounded; cells in 5 000 µg/mL putrescine group were in fragmentation without attachment. (2) The absorption values of cells in 0.5, 1.0, 5.0, 10.0, 50.0, 100.0, 500.0, 1 000.0 µg/mL putrescine groups, and control group were respectively 0.588 ± 0.055, 0.857 ± 0.031, 0.707 ± 0.031, 0.662 ± 0.023, 0.450 ± 0.019, 0.415 ± 0.014, 0.359 ± 0.020, 0.204 ± 0.030, and 0.447 ± 0.021, with statistically significant differences among them (χ(2) = 6.86, P = 0.009). The cell proliferation activity in 0.5, 1.0, 5.0, and 10.0 µg/mL putrescine groups was higher than that in control group (P0.05 or P0.01). The cell proliferation activity in 500.0 and 1 000.0 µg/mL putrescine groups was lower than that in control group (with P values below 0.01). The cell proliferation activity in 50.0 and 100.0 µg/mL putrescine groups was close to that in control group (with P values above 0.05). (3) There were statistically significant differences in the numbers of migrated cells between the putrescine groups and control group (F = 138.662, P0.001). The number of migrated cells was more in 1.0, 5.0, and 10.0 µg/mL putrescine groups than in control group (with P value below 0.01). The number of migrated cells was less in 500.0 and 1 000.0 µg/mL putrescine groups than in control group (with P value below 0.01). The number of migrated cells in 0.5, 50.0, and 100.0 µg/mL putrescine groups was close to that in control group (with P values above 0.05). (4) There were statistically significant differences in the apoptosis rate between the putrescine groups and control group (χ(2)=3.971, P=0.046). The cell apoptosis rate was lower in 0.5, 1.0, 5.0, and 10.0 µg/mL putrescine groups than in control group (with P values below 0.05). The cell apoptosis rate was higher in 500.0 and 1 000.0 µg/mL putrescine groups than in control group (with P values below 0.01). The cell apoptosis rates in 50.0 and 100.0 µg/mL putrescine groups were close to the cell apoptosis rate in control group (with P values above 0.05).Low concentration of putrescine can remarkably enhance the ability of proliferation and migration of HUVECs, while a high concentration of putrescine can obviously inhibit HUVECs proliferation and migration, and it induces apoptosis.

Co-substituted U-type hexagonal ferrite bulks, with composition of Ba4Ni2−xCoxFe36O60 (x=0.2, 0.4, 0.6, 0.8), were prepared by a conventional ceramic method. Saturation magnetization (4πMs), coercivity (Hc), and Curie temperature (Tc) were investigated. Anisotropy constant (K1) was calculated by fitting the magnetization curve (M-H) according to the law of approach to saturation, and anisotropy field (Ha) was calculated accordingly. The results reveal that all the samples possess the U-type hexagonal crystallographic structure. With increasing cobalt substitution content (x), the lattice parameters (a and c) almost remain the same owing to the similar radii of Ni2+ (0.72 A) Co2+ (0.74 A) ions. 4πMs goes up, while Hc Hc shows an opposite trend. K1 and Ha monotonously decrease resulting from that cobalt substitution weakens the c-axis orientation. Additionally, Tc increases from 467 °C to 484 °C.

Professional development offerings assist K–12 educators in addressing new and evolving classroom dynamics, circumstances, and situations. With the emerging demands of an increasingly science, technology, engineering, and mathematics (STEM)-infused society, teachers are challenged to provide high-quality service and equitable educational opportunities to all STEM education students, particularly to those students who traditionally are underrepresented in comparison to their peers in STEM education and/or have aspirations of participation in STEM-related careers. This study investigated K–12 STEM educator participation and perceived utility regarding professional development addressing specific needs of students with identified categorical disabilities and limited English proficiency (LEP). Collection and analysis methods employed data retrieval and tabulation from the 2011–2012 School and Staffing Survey (SASS) Teacher Questionnaire (TQ). The national restricted access dataset was used to identify targeted teacher populations as well as provide a profile of STEM teacher participation in practice-oriented professional development activities regarding the two specified student groups. The results were categorically summarized and compared across science, technology, and mathematics (STM) disciplines and also between STM educators, non-STM educators, and educators in general. The results indicated that STM teachers tended to engage in fewer professional development opportunities and dedicated fewer hours in the professional development regarding students with categorical disabilities and LEP than the remainder of the teaching population. Overall, STM teachers’ perceived utility of the provided professional development experience was lower than that of the remainder of the teaching population.

To explore the effects of different concentrations of putrescine on the proliferation, migration and apoptosis of human skin fibroblasts (HSF).HSF cultured in the presence of 0.5, 1.0, 5.0, 10, 50, 100, 500, and 1000 µg/ putrescine for 24 h were examined for the changes in the cell proliferation, migration, and apoptosis using MTS assay, Transwell migration assay, and flow cytometry, respectively.Compared with the control cells, HSF cultured with 0.5, 1.0, 5.0, and 10 µg/ putrescine showed significantly increased cell proliferation (P0.01), and the effect was the most obvious with 1 µg/ putrescine, whereas 500 and 1000 µg/ putrescine significantly reduced the cell proliferation (P0.01); 50 and 100 µg/ did not obviously affect the cell proliferation (P0.05). Putrescine at 1 µg/ most significantly enhanced the cell migration (P0.01), while at higher doses (50, 100, 500, and 1000 µg/) putrescine significantly suppressed the cell migration (P0.05); 0.5, 5.0, and 10 µg/ putrescine produced no obvious effects on the cell migration (P0.05). HSF treated with 0.5, 1.0, 5.0, and 10 µg/ putrescine obvious lowered the cell apoptosis rate compared with the control group (P0.01), and the cell apoptosis rate was the lowest in cells treated with 1 µg/ putrescine; but at the concentrations of 100, 500, and 1000 µg/, putrescine significantly increased the cell apoptosis rate (P0.01), while 50 µg/ml putrescine produced no obvious effect on cell apoptosis (P0.05).Low concentrations of putrescine can obviously enhance the proliferation ability and maintain normal migration ability of HSF in vitro, but at high concentrations, putrescine can obviously inhibit the cell migration and proliferation and induce cells apoptosis, suggesting the different roles of different concentrations of putrescine in wound healing.

In this paper, a new coding scheme for the multiple access channel (MAC) with noisy cooperative links is proposed. The cooperation cost is modelled by powers spent on exchanging common information at transmitters. The optimal power allocation policy is derived to explore the tradeoff between cooperation and transmission. For some important cases, optimal power allocation that maximizes weighted sum rate, is found analytically. The sufficient and necessary condition for which the sum and the individual rates are simultaneously maximized, is identified. Analytical and numerical results suggest that the transmitter, whose power budget is dominated by that of the other, acts purely as a relay. The cooperation gain becomes more significant when the difference between the power budgets is smaller.

Prototype learning is widely used in character recognition field. Unfortunately, current learning algorithms require intensive computation burden for large category applications, such as Japanese/Chinese character recognition. To resolve this challenge, a principled parallel method is proposed on GPUs instead of CPUs. We have implemented the method in mini-batch manner as well as stochastic gradient descent (SGD) manner. Our evaluations on a Chinese character database show that our method posses a high scalability while preserving its performance precision. Up to 194X speedup can be achieved in the case of mini-batch. Even to the more difficult SGD occasion, a more than 30-fold speedup is observed.

We consider a spectrum sharing protocol wherein the primary and secondary transmitters cooperatively relay each other's message. Transmission is done in two phases, with each transmitter attempting to decode messages from the other system transmission in a first phase. The second phase transmission consists of the decoded message superposed onto its own message. Priority is given to the primary system transmissions by having the primary message always transmitted over the two phases, while the secondary message is transmitted depending on successful decoding. We consider the scenario where the primary and secondary receivers are co-located, forming a virtual two-antenna receiver. We assess the performance of the system in terms of outage probability and characterize performance corresponding to each state of the Markov chain that governs the proposed transmission protocol. We show that joint decoding offers a 20 dB performance improvement over separate decoding for the primary user and 1.8 dB for the secondary user.

A new transmission protocol for joint cooperation between a primary and secondary user in a cognitive radio system is proposed. If successful decoding is achieved, the primary/secondary transmitters behave as relays for the other system. With more efficient use of the communication resource, performance is enhanced. The proposed scheme builds on the work in [1] and offers improved outage probabilities for both the primary and secondary systems. The paper provides a full characterization of the Markov Chain describing the protocol behavior and a closed form expression for the stationary distribution of the chain. Furthermore, exact computation or tight approximations for the outage probabilities for all states of the chain are derived. Simulation results exhibit a performance improvement between 22 and 48 percent for primary system and an uniform decrease of outage probability for secondary system.