arXiv subjects

Chunjie Luo

Publications and source records attributed to Chunjie Luo.

At least 37 records · Page 2Linked to original sources

arXiv2020

Finet: Using Fine-grained Batch Normalization to Train Light-weight Neural Networks

To build light-weight network, we propose a new normalization, Fine-grained Batch Normalization (FBN). Different from Batch Normalization (BN), which normalizes the final summation of the weighted inputs, FBN normalizes the intermediate state of the summation. We propose a novel light-weight network based on FBN, called Finet. At training time, the convolutional layer with FBN can be seen as an inverted bottleneck mechanism. FBN can be fused into convolution at inference time. After fusion, Finet uses the standard convolution with equal channel width, thus makes the inference more efficient. On ImageNet classification dataset, Finet achieves the state-of-art performance (65.706% accuracy with 43M FLOPs, and 73.786% accuracy with 303M FLOPs), Moreover, experiments show that Finet is more efficient than other state-of-art light-weight networks.

cs.LG↗

arXiv2020

Comparison and Benchmarking of AI Models and Frameworks on Mobile Devices

Due to increasing amounts of data and compute resources, deep learning achieves many successes in various domains. The application of deep learning on the mobile and embedded devices is taken more and more attentions, benchmarking and ranking the AI abilities of mobile and embedded devices becomes an urgent problem to be solved. Considering the model diversity and framework diversity, we propose a benchmark suite, AIoTBench, which focuses on the evaluation of the inference abilities of mobile and embedded devices. AIoTBench covers three typical heavy-weight networks: ResNet50, InceptionV3, DenseNet121, as well as three light-weight networks: SqueezeNet, MobileNetV2, MnasNet. Each network is implemented by three frameworks which are designed for mobile and embedded devices: Tensorflow Lite, Caffe2, Pytorch Mobile. To compare and rank the AI capabilities of the devices, we propose two unified metrics as the AI scores: Valid Images Per Second (VIPS) and Valid FLOPs Per Second (VOPS). Currently, we have compared and ranked 5 mobile devices using our benchmark. This list will be extended and updated soon after.

cs.LG↗

arXiv2020

Extended Batch Normalization

Batch normalization (BN) has become a standard technique for training the modern deep networks. However, its effectiveness diminishes when the batch size becomes smaller, since the batch statistics estimation becomes inaccurate. That hinders batch normalization's usage for 1) training larger model which requires small batches constrained by memory consumption, 2) training on mobile or embedded devices of which the memory resource is limited. In this paper, we propose a simple but effective method, called extended batch normalization (EBN). For NCHW format feature maps, extended batch normalization computes the mean along the (N, H, W) dimensions, as the same as batch normalization, to maintain the advantage of batch normalization. To alleviate the problem caused by small batch size, extended batch normalization computes the standard deviation along the (N, C, H, W) dimensions, thus enlarges the number of samples from which the standard deviation is computed. We compare extended batch normalization with batch normalization and group normalization on the datasets of MNIST, CIFAR-10/100, STL-10, and ImageNet, respectively. The experiments show that extended batch normalization alleviates the problem of batch normalization with small batch size while achieving close performances to batch normalization with large batch size.

cs.CV↗

arXiv2020

AIBench: An Agile Domain-specific Benchmarking Methodology and an AI Benchmark Suite

Domain-specific software and hardware co-design is encouraging as it is much easier to achieve efficiency for fewer tasks. Agile domain-specific benchmarking speeds up the process as it provides not only relevant design inputs but also relevant metrics, and tools. Unfortunately, modern workloads like Big data, AI, and Internet services dwarf the traditional one in terms of code size, deployment scale, and execution path, and hence raise serious benchmarking challenges. This paper proposes an agile domain-specific benchmarking methodology. Together with seventeen industry partners, we identify ten important end-to-end application scenarios, among which sixteen representative AI tasks are distilled as the AI component benchmarks. We propose the permutations of essential AI and non-AI component benchmarks as end-to-end benchmarks. An end-to-end benchmark is a distillation of the essential attributes of an industry-scale application. We design and implement a highly extensible, configurable, and flexible benchmark framework, on the basis of which, we propose the guideline for building end-to-end benchmarks, and present the first end-to-end Internet service AI benchmark. The preliminary evaluation shows the value of our benchmark suite---AIBench against MLPerf and TailBench for hardware and software designers, micro-architectural researchers, and code developers. The specifications, source code, testbed, and results are publicly available from the web site \url{http://www.benchcouncil.org/AIBench/index.html}.

cs.PF↗

arXiv2019

BOPS, Not FLOPS! A New Metric and Roofline Performance Model For Datacenter Computing

For emerging datacenter (in short, DC) workloads, such as online Internet services or offline data analytics, how to evaluate the upper bound performance and provide apple-to-apple comparisons are fundamental problems. To this end, a unified computation-centric metric is an essential requirement. As the most important computation-centric performance metric, FLOPS has guided computing systems evolutions for many years. However, our observations demonstrate that the average FLOPS efficiency of the DC workloads is only 0.1%, which implies that FLOPS is inappropriate for DC computing. To address the above issue, we propose BOPS (Basic Operations Per Second), which is the average number of BOPs (Basic OPerations) completed per second. We conduct the analysis on the characteristics of seventeen typical DC workloads and extract the minimum representative computation operations set, which is composed of integer and floating point computation operations of arithmetic, comparing and array addressing. Then, we propose the formalized BOPS definition and the BOPS based upper bound performance model. Finally, the BOPS measuring tool is also implemented. We perform experiments with seventeen DC workloads on three typical Intel processors platforms. First, BOPS can reflect the performance gap of different computing systems, the bias between the peak BOPS performance gap and the average DC workloads' wall clock time gap is no more than 10%. Second, the Sort workload can achieve 32% BOPS efficiency on the experimental platform. At last, we present two use cases of BOPS. One is the BOPS based system evaluation, we illustrate that BOPS can compare performance of workloads from multiple domains. The other is BOPS based optimizations. We show that under the guiding of the BOPS based upper bound model, the Sort workload and the Redis workload achieve 4.4X and 1.2X performance improvements respectively.

cs.PF↗

arXiv2019

HPC AI500: A Benchmark Suite for HPC AI Systems

In recent years, with the trend of applying deep learning (DL) in high performance scientific computing, the unique characteristics of emerging DL workloads in HPC raise great challenges in designing, implementing HPC AI systems. The community needs a new yard stick for evaluating the future HPC systems. In this paper, we propose HPC AI500 --- a benchmark suite for evaluating HPC systems that running scientific DL workloads. Covering the most representative scientific fields, each workload from HPC AI500 is based on real-world scientific DL applications. Currently, we choose 14 scientific DL benchmarks from perspectives of application scenarios, data sets, and software stack. We propose a set of metrics for comprehensively evaluating the HPC AI systems, considering both accuracy, performance as well as power and cost. We provide a scalable reference implementation of HPC AI500. HPC AI500 is a part of the open-source AIBench project, the specification and source code are publicly available from \url{http://www.benchcouncil.org/AIBench/index.html}.

cs.PF↗

arXiv2019

AIBench: An Industry Standard Internet Service AI Benchmark Suite

Today's Internet Services are undergoing fundamental changes and shifting to an intelligent computing era where AI is widely employed to augment services. In this context, many innovative AI algorithms, systems, and architectures are proposed, and thus the importance of benchmarking and evaluating them rises. However, modern Internet services adopt a microservice-based architecture and consist of various modules. The diversity of these modules and complexity of execution paths, the massive scale and complex hierarchy of datacenter infrastructure, the confidential issues of data sets and workloads pose great challenges to benchmarking. In this paper, we present the first industry-standard Internet service AI benchmark suite---AIBench with seventeen industry partners, including several top Internet service providers. AIBench provides a highly extensible, configurable, and flexible benchmark framework that contains loosely coupled modules. We identify sixteen prominent AI problem domains like learning to rank, each of which forms an AI component benchmark, from three most important Internet service domains: search engine, social network, and e-commerce, which is by far the most comprehensive AI benchmarking effort. On the basis of the AIBench framework, abstracting the real-world data sets and workloads from one of the top e-commerce providers, we design and implement the first end-to-end Internet service AI benchmark, which contains the primary modules in the critical paths of an industry scale application and is scalable to deploy on different cluster scales. The specifications, source code, and performance numbers are publicly available from the benchmark council web site http://www.benchcouncil.org/AIBench/index.html.

cs.CV↗

arXiv2018

Clustering Residential Electricity Load Curves via Community Detection in Network

Performing analytic of household load curves (LCs) has significant value in predicting individual electricity consumption patterns, and hence facilitate developing demand-response strategy, and finally achieve energy efficiency improvement and emission reduction. LC clustering is a widely used analytic technology, which discovers electricity consumption patterns by grouping similar LCs into same sub-groups. However, previous clustering methods treat each LC in the data set as an individual time series, ignoring the inherent relationship among different LCs, restraining the performance of the clustering methods. What's more, due to the significant volatility and uncertainty of LCs, the previous LC clustering approaches inevitably result in either lager number of clusters or huge variances within a cluster, which is unacceptable for actual application needs. In this paper, we proposed an integrated approach to address this issue. First, we converted the LC clustering task into the community detection task in network. Second, we proposed a clustering approach incorporated with community detection to improve the performance of LC clustering, which includes network construction, community detection and typical load profile extraction. The method convert the data set into a network in which the inherent relationship among LCs is represented by the edges of the network. Third, we construct a multi-layer typical load profile directory to make the trade-off between variances within a cluster and the number of the clusters, enabling the researchers to assess the LCs and the customers in different layers according to practical application requirements. The experiments demonstrate that our integrated approach outperform the state-of-the-art methods.

cs.DS↗

arXiv2018

BigDataBench: A Scalable and Unified Big Data and AI Benchmark Suite

Several fundamental changes in technology indicate domain-specific hardware and software co-design is the only path left. In this context, architecture, system, data management, and machine learning communities pay greater attention to innovative big data and AI algorithms, architecture, and systems. Unfortunately, complexity, diversity, frequently-changed workloads, and rapid evolution of big data and AI systems raise great challenges. First, the traditional benchmarking methodology that creates a new benchmark or proxy for every possible workload is not scalable, or even impossible for Big Data and AI benchmarking. Second, it is prohibitively expensive to tailor the architecture to characteristics of one or more application or even a domain of applications. We consider each big data and AI workload as a pipeline of one or more classes of units of computation performed on different initial or intermediate data inputs, each class of which we call a data motif. On the basis of our previous work that identifies eight data motifs taking up most of the run time of a wide variety of big data and AI workloads, we propose a scalable benchmarking methodology that uses the combination of one or more data motifs---to represent diversity of big data and AI workloads. Following this methodology, we present a unified big data and AI benchmark suite---BigDataBench 4.0, publicly available from~\url{http://prof.ict.ac.cn/BigDataBench}. This unified benchmark suite sheds new light on domain-specific hardware and software co-design: tailoring the system and architecture to characteristics of the unified eight data motifs other than one or more application case by case. Also, for the first time, we comprehensively characterize the CPU pipeline efficiency using the benchmarks of seven workload types in BigDataBench 4.0.

cs.DC↗

arXiv2018

Data Motif-based Proxy Benchmarks for Big Data and AI Workloads

For the architecture community, reasonable simulation time is a strong requirement in addition to performance data accuracy. However, emerging big data and AI workloads are too huge at binary size level and prohibitively expensive to run on cycle-accurate simulators. The concept of data motif, which is identified as a class of units of computation performed on initial or intermediate data, is the first step towards building proxy benchmark to mimic the real-world big data and AI workloads. However, there is no practical way to construct a proxy benchmark based on the data motifs to help simulation-based research. In this paper, we embark on a study to bridge the gap between data motif and a practical proxy benchmark. We propose a data motif-based proxy benchmark generating methodology by means of machine learning method, which combine data motifs with different weights to mimic the big data and AI workloads. Furthermore, we implement various data motifs using light-weight stacks and apply the methodology to five real-world workloads to construct a suite of proxy benchmarks, considering the data types, patterns, and distributions. The evaluation results show that our proxy benchmarks shorten the execution time by 100s times on real systems while maintaining the average system and micro-architecture performance data accuracy above 90%, even changing the input data sets or cluster configurations. Moreover, the generated proxy benchmarks reflect consistent performance trends across different architectures. To facilitate the community, we will release the proxy benchmarks on the project homepage http://prof.ict.ac.cn/BigDataBench.

cs.DC↗

arXiv2018

Data Motifs: A Lens Towards Fully Understanding Big Data and AI Workloads

The complexity and diversity of big data and AI workloads make understanding them difficult and challenging. This paper proposes a new approach to modelling and characterizing big data and AI workloads. We consider each big data and AI workload as a pipeline of one or more classes of units of computation performed on different initial or intermediate data inputs. Each class of unit of computation captures the common requirements while being reasonably divorced from individual implementations, and hence we call it a data motif. For the first time, among a wide variety of big data and AI workloads, we identify eight data motifs that take up most of the run time of those workloads, including Matrix, Sampling, Logic, Transform, Set, Graph, Sort and Statistic. We implement the eight data motifs on different software stacks as the micro benchmarks of an open-source big data and AI benchmark suite ---BigDataBench 4.0 (publicly available from http://prof.ict.ac.cn/BigDataBench), and perform comprehensive characterization of those data motifs from perspective of data sizes, types, sources, and patterns as a lens towards fully understanding big data and AI workloads. We believe the eight data motifs are promising abstractions and tools for not only big data and AI benchmarking, but also domain-specific hardware and software co-design.

cs.DC↗

arXiv2018

Data Dwarfs: A Lens Towards Fully Understanding Big Data and AI Workloads

The complexity and diversity of big data and AI workloads make understanding them difficult and challenging. This paper proposes a new approach to characterizing big data and AI workloads. We consider each big data and AI workload as a pipeline of one or more classes of unit of computations performed on different initial or intermediate data inputs. Each class of unit of computation captures the common requirements while being reasonably divorced from individual implementations, and hence we call it a data dwarf. For the first time, among a wide variety of big data and AI workloads, we identify eight data dwarfs that takes up most of run time, including Matrix, Sampling, Logic, Transform, Set, Graph, Sort and Statistic. We implement the eight data dwarfs on different software stacks as the micro benchmarks of an open-source big data and AI benchmark suite, and perform comprehensive characterization of those data dwarfs from perspective of data sizes, types, sources, and patterns as a lens towards fully understanding big data and AI workloads.

cs.DC↗

arXiv2018

A veracity preserving model for synthesizing scalable electricity load profiles

Electricity users are the major players of the electric systems, and electricity consumption is growing at an extraordinary rate. The research on electricity consumption behaviors is becoming increasingly important to design and deployment of the electric systems. Unfortunately, electricity load profiles are difficult to acquire. Data synthesis is one of the best approaches to solving the lack of data, and the key is the model that preserves the real electricity consumption behaviors. In this paper, we propose a hierarchical multi-matrices Markov Chain (HMMC) model to synthesize scalable electricity load profiles that preserve the real consumption behavior on three time scales: per day, per week, and per year. To promote the research on the electricity consumption behavior, we use the HMMC approach to model two distinctive raw electricity load profiles. One is collected from the resident sector, and the other is collected from the non-resident sectors, including different industries such as education, finance, and manufacturing. The experiments show our model performs much better than the classical Markov Chain model. We publish two trained models online, and researchers can directly use these trained models to synthesize scalable electricity load profiles for further researches.

cs.OH↗

arXiv2017

HybridTune: Spatio-temporal Data and Model Driven Performance Diagnosis for Big Data Systems

With tremendous growing interests in Big Data systems, analyzing and facilitating their performance improvement become increasingly important. Although there have much research efforts for improving Big Data systems performance, efficiently analysing and diagnosing performance bottlenecks over these massively distributed systems remain a major challenge. In this paper, we propose a spatio-temporal correlation analysis approach based on stage characteristic and distribution characteristic of Big Data applications, which can associate the multi-level performance data fine-grained. On the basis of correlation data, we define some priori rules, select features and vectorize the corresponding datasets for different performance bottlenecks, such as, workload imbalance, data skew, abnormal node and outlier metrics. And then, we utilize the data and model driven algorithms for bottlenecks detection and diagnosis. In addition, we design and develop a lightweight, extensible tool HybridTune, and validate the diagnosis effectiveness of our tool with BigDataBench on several benchmark experiments in which the outperform state-of-the-art methods. Our experiments show that the accuracy of abnormal/outlier detection we obtained reaches about 80%. At last, we report several Spark and Hadoop use cases, which are demonstrated how HybridTune supports users to carry out the performance analysis and diagnosis efficiently on the Spark and Hadoop applications, and our experiences demonstrate HybridTune can help users find the performance bottlenecks and provide optimization recommendations.

cs.DC↗

arXiv2017

A Dwarf-based Scalable Big Data Benchmarking Methodology

Different from the traditional benchmarking methodology that creates a new benchmark or proxy for every possible workload, this paper presents a scalable big data benchmarking methodology. Among a wide variety of big data analytics workloads, we identify eight big data dwarfs, each of which captures the common requirements of each class of unit of computation while being reasonably divorced from individual implementations. We implement the eight dwarfs on different software stacks, e.g., OpenMP, MPI, Hadoop as the dwarf components. For the purpose of architecture simulation, we construct and tune big data proxy benchmarks using the directed acyclic graph (DAG)-like combinations of the dwarf components with different weights to mimic the benchmarks in BigDataBench. Our proxy benchmarks preserve the micro-architecture, memory, and I/O characteristics, and they shorten the simulation time by 100s times while maintain the average micro-architectural data accuracy above 90 percentage on both X86 64 and ARMv8 processors. We will open-source the big data dwarf components and proxy benchmarks soon.

cs.AR↗

arXiv2017

Cosine Normalization: Using Cosine Similarity Instead of Dot Product in Neural Networks

Traditionally, multi-layer neural networks use dot product between the output vector of previous layer and the incoming weight vector as the input to activation function. The result of dot product is unbounded, thus increases the risk of large variance. Large variance of neuron makes the model sensitive to the change of input distribution, thus results in poor generalization, and aggravates the internal covariate shift which slows down the training. To bound dot product and decrease the variance, we propose to use cosine similarity or centered cosine similarity (Pearson Correlation Coefficient) instead of dot product in neural networks, which we call cosine normalization. We compare cosine normalization with batch, weight and layer normalization in fully-connected neural networks as well as convolutional networks on the data sets of MNIST, 20NEWS GROUP, CIFAR-10/100 and SVHN. Experiments show that cosine normalization achieves better performance than other normalization techniques.

cs.LG↗

arXiv2015

Identifying Dwarfs Workloads in Big Data Analytics

Big data benchmarking is particularly important and provides applicable yardsticks for evaluating booming big data systems. However, wide coverage and great complexity of big data computing impose big challenges on big data benchmarking. How can we construct a benchmark suite using a minimum set of units of computation to represent diversity of big data analytics workloads? Big data dwarfs are abstractions of extracting frequently appearing operations in big data computing. One dwarf represents one unit of computation, and big data workloads are decomposed into one or more dwarfs. Furthermore, dwarfs workloads rather than vast real workloads are more cost-efficient and representative to evaluate big data systems. In this paper, we extensively investigate six most important or emerging application domains i.e. search engine, social network, e-commerce, multimedia, bioinformatics and astronomy. After analyzing forty representative algorithms, we single out eight dwarfs workloads in big data analytics other than OLAP, which are linear algebra, sampling, logic operations, transform operations, set operations, graph operations, statistic operations and sort.

cs.DB↗

arXiv2015

Understanding Big Data Analytic Workloads on Modern Processors

Big data analytics applications play a significant role in data centers, and hence it has become increasingly important to understand their behaviors in order to further improve the performance of data center computer systems, in which characterizing representative workloads is a key practical problem. In this paper, after investigating three most impor- tant application domains in terms of page views and daily visitors, we chose 11 repre- sentative data analytics workloads and characterized their micro-architectural behaviors by using hardware performance counters, so as to understand the impacts and implications of data analytics workloads on the systems equipped with modern superscalar out-of-order processors. Our study reveals that big data analytics applications themselves share many inherent characteristics, which place them in a different class from traditional workloads and scale-out services. To further understand the characteristics of big data analytics work- loads we performed a correlation analysis of CPI (cycles per instruction) with other micro- architecture level characteristics and an investigation of the big data software stack impacts on application behaviors. Our correlation analysis showed that even though big data ana- lytics workloads own notable pipeline front end stalls, the main factors affecting the CPI performance are long latency data accesses rather than the front end stalls. Our software stack investigation found that the typical big data software stack significantly contributes to the front end stalls and incurs bigger working set. Finally we gave several recommen- dations for architects, programmers and big data system designers with the knowledge acquired from this paper.

cs.DC↗