arXiv subjects

Fei Tang

Publications and source records attributed to Fei Tang.

53 records · Page 3Linked to original sources

arXiv2022

Traffic4cast at NeurIPS 2021 -- Temporal and Spatial Few-Shot Transfer Learning in Gridded Geo-Spatial Processes

The IARAI Traffic4cast competitions at NeurIPS 2019 and 2020 showed that neural networks can successfully predict future traffic conditions 1 hour into the future on simply aggregated GPS probe data in time and space bins. We thus reinterpreted the challenge of forecasting traffic conditions as a movie completion task. U-Nets proved to be the winning architecture, demonstrating an ability to extract relevant features in this complex real-world geo-spatial process. Building on the previous competitions, Traffic4cast 2021 now focuses on the question of model robustness and generalizability across time and space. Moving from one city to an entirely different city, or moving from pre-COVID times to times after COVID hit the world thus introduces a clear domain shift. We thus, for the first time, release data featuring such domain shifts. The competition now covers ten cities over 2 years, providing data compiled from over 10^12 GPS probe data. Winning solutions captured traffic dynamics sufficiently well to even cope with these complex domain shifts. Surprisingly, this seemed to require only the previous 1h traffic dynamic history and static road graph as input.

cs.LG↗

arXiv2021

AIBench Scenario: Scenario-distilling AI Benchmarking

Modern real-world application scenarios like Internet services consist of a diversity of AI and non-AI modules with huge code sizes and long and complicated execution paths, which raises serious benchmarking or evaluating challenges. Using AI components or micro benchmarks alone can lead to error-prone conclusions. This paper presents a methodology to attack the above challenge. We formalize a real-world application scenario as a Directed Acyclic Graph-based model and propose the rules to distill it into a permutation of essential AI and non-AI tasks, which we call a scenario benchmark. Together with seventeen industry partners, we extract nine typical scenario benchmarks. We design and implement an extensible, configurable, and flexible benchmark framework. We implement two Internet service AI scenario benchmarks based on the framework as proxies to two real-world application scenarios. We consider scenario, component, and micro benchmarks as three indispensable parts for evaluating. Our evaluation shows the advantage of our methodology against using component or micro AI benchmarks alone. The specifications, source code, testbed, and results are publicly available from \url{https://www.benchcouncil.org/aibench/scenario/}.

cs.PF↗

arXiv2021

A remark on a paper of Krotov and Hopfield [arXiv:2008.06996]

In their recent paper titled "Large Associative Memory Problem in Neurobiology and Machine Learning" [arXiv:2008.06996] the authors gave a biologically plausible microscopic theory from which one can recover many dense associative memory models discussed in the literature. We show that the layers of the recent "MLP-mixer" [arXiv:2105.01601] as well as the essentially equivalent model in [arXiv:2105.02723] are amongst them.

q-bio.NC↗

arXiv2021

AIBench Training: Balanced Industry-Standard AI Training Benchmarking

Earlier-stage evaluations of a new AI architecture/system need affordable benchmarks. Only using a few AI component benchmarks like MLPerfalone in the other stages may lead to misleading conclusions. Moreover, the learning dynamics are not well understood, and the benchmarks' shelf-life is short. This paper proposes a balanced benchmarking methodology. We use real-world benchmarks to cover the factors space that impacts the learning dynamics to the most considerable extent. After performing an exhaustive survey on Internet service AI domains, we identify and implement nineteen representative AI tasks with state-of-the-art models. For repeatable performance ranking (RPR subset) and workload characterization (WC subset), we keep two subsets to a minimum for affordability. We contribute by far the most comprehensive AI training benchmark suite. The evaluations show: (1) AIBench Training (v1.1) outperforms MLPerfTraining (v0.7) in terms of diversity and representativeness of model complexity, computational cost, convergent rate, computation, and memory access patterns, and hotspot functions; (2) Against the AIBench full benchmarks, its RPR subset shortens the benchmarking cost by 64%, while maintaining the primary workload characteristics; (3) The performance ranking shows the single-purpose AI accelerator like TPU with the optimized TensorFlowframework performs better than that of GPUs while losing the latter's general support for various AI models. The specification, source code, and performance numbers are available from the AIBench homepage https://www.benchcouncil.org/aibench-training/index.html.

cs.AI↗

arXiv2021

HPC AI500: Representative, Repeatable and Simple HPC AI Benchmarking

Recent years witness a trend of applying large-scale distributed deep learning algorithms (HPC AI) in both business and scientific computing areas, whose goal is to speed up the training time to achieve a state-of-the-art quality. The HPC AI benchmarks accelerate the process. Unfortunately, benchmarking HPC AI systems at scale raises serious challenges. This paper presents a representative, repeatable and simple HPC AI benchmarking methodology. Among the seventeen AI workloads of AIBench Training -- by far the most comprehensive AI Training benchmarks suite -- we choose two representative and repeatable AI workloads. The selected HPC AI benchmarks include both business and scientific computing: Image Classification and Extreme Weather Analytics. To rank HPC AI systems, we present a new metric named Valid FLOPS, emphasizing both throughput performance and a target quality. The specification, source code, datasets, and HPC AI500 ranking numbers are publicly available from \url{https://www.benchcouncil.org/HPCAI500/}.

cs.PF↗

arXiv2020

HPC AI500: The Methodology, Tools, Roofline Performance Models, and Metrics for Benchmarking HPC AI Systems

The recent years witness a trend of applying large-scale distributed deep learning in both business and scientific computing areas, whose goal is to speed up the training time to achieve a state-of-the-art quality. The HPC community feels a great interest in building the HPC AI systems that are dedicated to running those workloads. The HPC AI benchmarks accelerate the process. Unfortunately, benchmarking HPC AI systems at scale raises serious challenges. None of previous HPC AI benchmarks achieve the goal of being equivalent, relevant, representative, affordable, and repeatable. This paper presents a comprehensive methodology, tools, Roofline performance models, and innovative metrics for benchmarking, optimizing, and ranking HPC AI systems, which we call HPC AI500 V2.0. We abstract the HPC AI system into nine independent layers, and present explicit benchmarking rules and procedures to assure equivalence of each layer, repeatability, and replicability. On the basis of AIBench -- by far the most comprehensive AI benchmarks suite, we present and build two HPC AI benchmarks from both business and scientific computing: Image Classification, and Extreme Weather Analytics, achieving both representativeness and affordability. To rank the performance and energy-efficiency of HPC AI systems, we propose Valid FLOPS, and Valid FLOPS per watt, which impose a penalty on failing to achieve the target quality. We propose using convolution and GEMM -- the two most intensively-used kernel functions to measure the upper bound performance of the HPC AI systems, and present HPC AI roofline models for guiding performance optimizations. The evaluations show our methodology, benchmarks, performance models, and metrics can measure, optimize, and rank the HPC AI systems in a scalable, simple, and affordable way. HPC AI500 V2.0 are publicly available from http://www.benchcouncil.org/benchhub/hpc-ai500-benchmark.

cs.PF↗

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

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

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↗

arXiv2017

Isolate First, Then Share: a New OS Architecture for Datacenter Computing

This paper presents the "isolate first, then share" OS model in which the processor cores, memory, and devices are divided up between disparate OS instances and a new abstraction, subOS, is proposed to encapsulate an OS instance that can be created, destroyed, and resized on-the-fly. The intuition is that this avoids shared kernel states between applications, which in turn reduces performance loss caused by contention. We decompose the OS into the supervisor and several subOSes running at the same privilege level: a subOS directly manages physical resources, while the supervisor can create, destroy, resize a subOS on-the-fly. The supervisor and subOSes have few state sharing, but fast inter-subOS communication mechanisms are provided on demand. We present the first implementation, RainForest, which supports unmodified Linux binaries. Our comprehensive evaluation shows RainForest outperforms Linux with four different kernels, LXC, and Xen in terms of worst-case and average performance most of time when running a large number of benchmarks. The source code is available soon.

cs.OS↗

arXiv2017

Random Forest Missing Data Algorithms

Random forest (RF) missing data algorithms are an attractive approach for dealing with missing data. They have the desirable properties of being able to handle mixed types of missing data, they are adaptive to interactions and nonlinearity, and they have the potential to scale to big data settings. Currently there are many different RF imputation algorithms but relatively little guidance about their efficacy, which motivated us to study their performance. Using a large, diverse collection of data sets, performance of various RF algorithms was assessed under different missing data mechanisms. Algorithms included proximity imputation, on the fly imputation, and imputation utilizing multivariate unsupervised and supervised splitting---the latter class representing a generalization of a new promising imputation algorithm called missForest. Performance of algorithms was assessed by ability to impute data accurately. Our findings reveal RF imputation to be generally robust with performance improving with increasing correlation. Performance was good under moderate to high missingness, and even (in certain cases) when data was missing not at random.

stat.ML↗

arXiv2016

AccuracyTrader: Accuracy-aware Approximate Processing for Low Tail Latency and High Result Accuracy in Cloud Online Services

Modern latency-critical online services such as search engines often process requests by consulting large input data spanning massive parallel components. Hence the tail latency of these components determines the service latency. To trade off result accuracy for tail latency reduction, existing techniques use the components responding before a specified deadline to produce approximate results. However, they may skip a large proportion of components when load gets heavier, thus incurring large accuracy losses. This paper presents AccuracyTrader that produces approximate results with small accuracy losses while maintaining low tail latency. AccuracyTrader aggregates information of input data on each component to create a small synopsis, thus enabling all components producing initial results quickly using their synopses. AccuracyTrader also uses synopses to identify the parts of input data most related to arbitrary requests' result accuracy, thus first using these parts to improve the produced results in order to minimize accuracy losses. We evaluated AccuracyTrader using workloads in real services. The results show: (i) AccuracyTrader reduces tail latency by over 40 times with accuracy losses of less than 7% compared to existing exact processing techniques; (ii) when using the same latency, AccuracyTrader reduces accuracy losses by over 13 times comparing to existing approximate processing techniques.

cs.DC↗

arXiv2015

Coupled Cluster Treatment of the Alternating Bond Diamond Chain

By the analytical coupled cluster method (CCM), we study both the ground state and lowest-lying excited-state properties of the alternating bond diamond chain. The numerical exact diagonalization (ED) method is also applied to the chain to verify the accuracy of CCM results. The ED results show that the ground-state phase diagram contains two exact spin cluster solid ground states, namely, the tetramer-dimer (TD) state and dimer state, and the ferrimagnetic long-range-ordered state. We prove that the two exact spin cluster solid ground states can both be formed by CCM. Moreover, the exact spin gap in the TD state can be obtained by CCM. In the ferrimagnetic region, we find that the CCM results for some physical quantities, such as the ground-state energy, the sublattice magnetizations, and the antiferromagnetic gap, are comparable to the results obtained by numerical methods. The critical line dividing the TD state from the ferrimagnetic state is also given by CCM and is in perfect agreement with that determined by the ED method.

cond-mat.str-el↗

arXiv2015

Coupled cluster treatment of one quasi-one-dimensional coupled spin triangles

By using the coupled cluster method (CCM) and the numerical exact diagonalization (ED) method, we investigated the properties of the one quasi-one-dimensional coupled spin triangles. The results of ED disclose that the system is in the exact tetramer-dimer (TD) state if a ac1, we find that CCM results for some physical quantities, such as the ground state energy, the magnetization and the antiferromagnetic gap are in excellent agreement with the results obtained by ED. Thus, CCM can be used to accurately analyze the properties of the quasi-one-dimensional coupled spin triangles in the whole parameter region. We believe that it can also be applied to investigating the properties of other quasi-one spin systems reliably.

cond-mat.str-el↗

arXiv2015

Analytical and numerical studies of the one-dimensional sawtooth chain

By using the analytical coupled cluster method, the numerical exact diagonalization method, and the numerical density matrix renormalization group method, we investigated the properties of the one-dimensional sawtooth chain. The results of the coupled cluster method based on Neel state demonstrate that the ground state is in the quasi-Neel-long-range order state when a ac2. The results of the coupled cluster method and the density matrix renormalization group method both disclose that the type of the quantum phase transition occurring at ac2 belongs to the first-order transition.

cond-mat.str-el↗

arXiv2014

Frustration induced noncollinear magnetic order phase in one-dimensional Heisenberg chain with alternating antiferromagnetic and ferromagnetic next nearest neighbor interactions

By using the coupled cluster method, the numerical exact diagonalization method, and the numerical density matrix renormalization group method, we investigated the properties of the one-dimensional Heisenberg chain with alternating antiferromagnetic and ferromagnetic next nearest neighbor interactions. In the classical limit, the ground state is in the collinear Neel state if a<1/2, while for a>1/2, there is an noncollinear canted state. For the quantum case, we found that, although the classical Neel state is absent, the canted state exists if the frustration parameter a exceeds a critical point ac1. The precise critical point ac1 can be determined by using the coupled cluster method and the numerical exact diagonalization method separately. The results of the coupled cluster method and the exact diagonalization method both disclose that the type of phase transition occurring at ac1 changes from a classical second-order transition to a quantum first-order transition due to quantum fluctuation. Although there is another critical point ac2 in a finite system at which the ground state evolves from the canted state to the collinear Neel plus ferromagnetic state, that state is absent because ac2 tends to infinity in the thermodynamic limit.

cond-mat.str-el↗