arXiv subjects

Mao Yang

Publications and source records attributed to Mao Yang.

At least 73 records · Page 4Linked to original sources

arXiv2023

SpaceEvo: Hardware-Friendly Search Space Design for Efficient INT8 Inference

The combination of Neural Architecture Search (NAS) and quantization has proven successful in automatically designing low-FLOPs INT8 quantized neural networks (QNN). However, directly applying NAS to design accurate QNN models that achieve low latency on real-world devices leads to inferior performance. In this work, we find that the poor INT8 latency is due to the quantization-unfriendly issue: the operator and configuration (e.g., channel width) choices in prior art search spaces lead to diverse quantization efficiency and can slow down the INT8 inference speed. To address this challenge, we propose SpaceEvo, an automatic method for designing a dedicated, quantization-friendly search space for each target hardware. The key idea of SpaceEvo is to automatically search hardware-preferred operators and configurations to construct the search space, guided by a metric called Q-T score to quantify how quantization-friendly a candidate search space is. We further train a quantized-for-all supernet over our discovered search space, enabling the searched models to be directly deployed without extra retraining or quantization. Our discovered models establish new SOTA INT8 quantized accuracy under various latency constraints, achieving up to 10.1% accuracy improvement on ImageNet than prior art CNNs under the same latency. Extensive experiments on diverse edge devices demonstrate that SpaceEvo consistently outperforms existing manually-designed search spaces with up to 2.5x faster speed while achieving the same accuracy.

cs.CV↗

arXiv2023

Gamify Stencil Dwarf on Cloud for Democratizing Scientific Computing

Stencil computation is one of the most important kernels in various scientific computing. Nowadays, most Stencil-driven scientific computing still relies heavily on supercomputers, suffering from expensive access, poor scalability, and duplicated optimizations. This paper proposes Tetris, the first system for high-performance Stencil on heterogeneous CPU+GPU, towards democratizing Stencil-driven scientific computing on Cloud. In Tetris, polymorphic tiling tetrominoes are first proposed to bridge different hardware architectures and various application contexts with a perfect spatial and temporal tessellation automatically. Tetris is contributed by three main components: (1) Underlying hardware characteristics are first captured to achieve a sophisticated Pattern Mapping by register-level tetrominoes; (2) An efficient Locality Enhancer is first presented for data reuse on spatial and temporal dimensions simultaneously by cache/SMEM-level tetrominoes; (3) A novel Concurrent Scheduler is first designed to exploit the full potential of on-cloud memory and computing power by memory-level tetrominoes. Tetris is orthogonal to (and complements) the optimizations or deployments for a wide variety of emerging and legacy scientific computing applications. Results of thermal diffusion simulation demonstrate that the performance is improved by 29.6x, reducing time cost from day to hour, while preserving the original accuracy.

cs.DC↗

arXiv2023

Insights on the Next Generation WLAN: High Experiences (HEX)

Wireless local area network (WLAN) witnesses a very fast growth in the past 20 years by taking the maximum throughput as the key technical objective. However, the quality of experience (QoE) is the most important concern of wireless network users. In this article, we point out that poor QoE is the most challenging problem of the current WLAN, and further analyze the key technical problems that cause the poor QoE of WLAN, including fully distributed networking architecture, chaotic random access, awkward ``high capability'', coarse-grained QoS architecture, ubiquitous and complicated interference, ``no place'' for artificial intelligence (AI), and heavy burden of standard evolving. To the best of our knowledge, this is the first work to point out that poor QoE is the most challenging problem of the current WLAN, and the first work to systematically analyze the technical problems that cause the poor QoE of WLAN. We highly suggest that achieving high experiences (HEX) be the key objective of the next generation WLAN.

cs.NI↗

arXiv2023

A Neural Corpus Indexer for Document Retrieval

Current state-of-the-art document retrieval solutions mainly follow an index-retrieve paradigm, where the index is hard to be directly optimized for the final retrieval target. In this paper, we aim to show that an end-to-end deep neural network unifying training and indexing stages can significantly improve the recall performance of traditional methods. To this end, we propose Neural Corpus Indexer (NCI), a sequence-to-sequence network that generates relevant document identifiers directly for a designated query. To optimize the recall performance of NCI, we invent a prefix-aware weight-adaptive decoder architecture, and leverage tailored techniques including query generation, semantic document identifiers, and consistency-based regularization. Empirical studies demonstrated the superiority of NCI on two commonly used academic benchmarks, achieving +21.4% and +16.8% relative enhancement for Recall@1 on NQ320k dataset and R-Precision on TriviaQA dataset, respectively, compared to the best baseline method.

cs.IR↗

arXiv2023

Sensing Assisted Communication for the Next Generation mmWave WLAN: System Simulation Perspective

With the proliferating of wireless demands, wireless local area network (WLAN) becomes one of the most important wireless networks. Network intelligence is promising for the next generation wireless networks, captured lots of attentions. Sensing is one efficient enabler to achieve network intelligence since utilizing sensing can obtain diverse and valuable non-communication information. Thus, integrating sensing and communications (ISAC) is a promising technology for future wireless networks. Sensing assisted communication (SAC) is an important branch of ISAC, but there are few related works focusing on the systematical and comprehensive analysis on SAC in WLAN. This article is the first work to systematically analyze SAC in the next generation WLAN from the system simulation perspective. We analyze the scenarios and advantages of SAC. Then, from system simulation perspective, several sources of performance gain brought from SAC are proposed, i.e. beam link failure, protocol overhead, and intra-physical layer protocol data unit (intra-PPDU) performance decrease, while several important influencing factors are described in detail. Performance evaluation is deeply analyzed and the performance gain of the SAC in both living room and street canyon scenarios are verified by system simulation. Finally, we provide our insights on the future directions of SAC for the next generation WLAN.

cs.NI↗

arXiv2023

SuperScaler: Supporting Flexible DNN Parallelization via a Unified Abstraction

With the growing model size, deep neural networks (DNN) are increasingly trained over massive GPU accelerators, which demands a proper parallelization plan that transforms a DNN model into fine-grained tasks and then schedules them to GPUs for execution. Due to the large search space, the contemporary parallelization plan generators often rely on empirical rules that couple transformation and scheduling, and fall short in exploring more flexible schedules that yield better memory usage and compute efficiency. This tension can be exacerbated by the emerging models with increasing complexity in their structure and model size. SuperScaler is a system that facilitates the design and generation of highly flexible parallelization plans. It formulates the plan design and generation into three sequential phases explicitly: model transformation, space-time scheduling, and data dependency preserving. Such a principled approach decouples multiple seemingly intertwined factors and enables the composition of highly flexible parallelization plans. As a result, SuperScaler can not only generate empirical parallelization plans, but also construct new plans that achieve up to 3.5X speedup compared to state-of-the-art solutions like DeepSpeed, Megatron and Alpa, for emerging DNN models like Swin-Transformer and AlphaFold2, as well as well-optimized models like GPT-3.

cs.DC↗

arXiv2022

Novel Approach to Structural Relaxation of Materials in Optically Excited States

We present a first-principles method for relaxing a material's geometry in an optically excited state. This method, based on the Bethe-Salpeter equation, consists of solving coupled equations for exciton wavefunctions and atomic displacements. Our approach allows for structural relaxation of excited states to be achieved through a single iteration. As results, one obtains not only energy and wavefunction of the thus modified, i.e. self-trapped, exciton, but also the mechanism of relaxation in terms of atomic displacements in the respective phonon eigenmodes. We demonstrate and evaluate our formalism with the example of the three molecules CO, H$_{2}$O, and NH$_{3}$.

physics.comp-ph↗

arXiv2022

On the necessity of separating MAC protocols into data and control planes

The network protocol architecture not only can be designed from the traditional view of layers, but also can be designed from the view of planes, i.e., the data, control and management planes. Media access control (MAC) is a function of the data link layer, and thus the MAC protocols involve of both the data and control planes. However, although the international wireless local area network (WLAN) standard, IEEE 802.11 or Wi-Fi, has developed over 20 years, the control plane of the MAC protocols is not explicitly described yet. Thus, does it need to separate the MAC protocols into data and control planes? If not, are there some problems in existing hybrid architecture? To answer above questions, we analyse the possible problems of the current MAC protocols in IEEE 802.11, particularly in std 802.11-2020. These problems can be seen as new starts for the next study of the WLAN for the next generation (WNG).

cs.NI↗

arXiv2022

SwiftPruner: Reinforced Evolutionary Pruning for Efficient Ad Relevance

Ad relevance modeling plays a critical role in online advertising systems including Microsoft Bing. To leverage powerful transformers like BERT in this low-latency setting, many existing approaches perform ad-side computations offline. While efficient, these approaches are unable to serve cold start ads, resulting in poor relevance predictions for such ads. This work aims to design a new, low-latency BERT via structured pruning to empower real-time online inference for cold start ads relevance on a CPU platform. Our challenge is that previous methods typically prune all layers of the transformer to a high, uniform sparsity, thereby producing models which cannot achieve satisfactory inference speed with an acceptable accuracy. In this paper, we propose SwiftPruner - an efficient framework that leverages evolution-based search to automatically find the best-performing layer-wise sparse BERT model under the desired latency constraint. Different from existing evolution algorithms that conduct random mutations, we propose a reinforced mutator with a latency-aware multi-objective reward to conduct better mutations for efficiently searching the large space of layer-wise sparse models. Extensive experiments demonstrate that our method consistently achieves higher ROC AUC and lower latency than the uniform sparse baseline and state-of-the-art search methods. Remarkably, under our latency requirement of 1900us on CPU, SwiftPruner achieves a 0.86% higher AUC than the state-of-the-art uniform sparse baseline for BERT-Mini on a large scale real-world dataset. Online A/B testing shows that our model also achieves a significant 11.7% cut in the ratio of defective cold start ads with satisfactory real-time serving latency.

cs.IR↗

arXiv2021

SPANN: Highly-efficient Billion-scale Approximate Nearest Neighbor Search

The in-memory algorithms for approximate nearest neighbor search (ANNS) have achieved great success for fast high-recall search, but are extremely expensive when handling very large scale database. Thus, there is an increasing request for the hybrid ANNS solutions with small memory and inexpensive solid-state drive (SSD). In this paper, we present a simple but efficient memory-disk hybrid indexing and search system, named SPANN, that follows the inverted index methodology. It stores the centroid points of the posting lists in the memory and the large posting lists in the disk. We guarantee both disk-access efficiency (low latency) and high recall by effectively reducing the disk-access number and retrieving high-quality posting lists. In the index-building stage, we adopt a hierarchical balanced clustering algorithm to balance the length of posting lists and augment the posting list by adding the points in the closure of the corresponding clusters. In the search stage, we use a query-aware scheme to dynamically prune the access of unnecessary posting lists. Experiment results demonstrate that SPANN is 2$\times$ faster than the state-of-the-art ANNS solution DiskANN to reach the same recall quality $90\%$ with same memory cost in three billion-scale datasets. It can reach $90\%$ recall@1 and recall@10 in just around one millisecond with only 32GB memory cost. Code is available at: {\footnotesize\color{blue}{\url{https://github.com/microsoft/SPTAG}}}.

cs.DB↗

arXiv2021

WRENCH: A Comprehensive Benchmark for Weak Supervision

Recent Weak Supervision (WS) approaches have had widespread success in easing the bottleneck of labeling training data for machine learning by synthesizing labels from multiple potentially noisy supervision sources. However, proper measurement and analysis of these approaches remain a challenge. First, datasets used in existing works are often private and/or custom, limiting standardization. Second, WS datasets with the same name and base data often vary in terms of the labels and weak supervision sources used, a significant "hidden" source of evaluation variance. Finally, WS studies often diverge in terms of the evaluation protocol and ablations used. To address these problems, we introduce a benchmark platform, WRENCH, for thorough and standardized evaluation of WS approaches. It consists of 22 varied real-world datasets for classification and sequence tagging; a range of real, synthetic, and procedurally-generated weak supervision sources; and a modular, extensible framework for WS evaluation, including implementations for popular WS methods. We use WRENCH to conduct extensive comparisons over more than 120 method variants to demonstrate its efficacy as a benchmark platform. The code is available at https://github.com/JieyuZ2/wrench.

cs.LG↗

arXiv2021

Structure-Composition-Property Relationships in Antiperovskite Nitrides: Guiding a Rational Alloy Design

The alloy strategy through A- or X-site is a common method for experimental preparation of high-performance and stable lead-based perovskite solar cells. As one of the important candidates for lead-free and stable photovoltaic absorber, the inorganic antiperovskite family has recently been reported to exhibit excellent optoelectronic properties. However, the current reports on the design of antiperovskite alloys are rare. In this work, we investigated the previously overlooked electronic property (e.g., conduction band convergence), static dielectric constant, and exciton binding energy in inorganic antiperovskite nitrides by first-principles calculations. Then, we reveal a linear relationship between tolerance factor and various physical quantities. Guided by the established structure-composition-property relationship in six antiperovskite nitrides X3NA (X2+ = Mg2+, Ca2+, Sr2+; A3- = P3-, As3-, Sb3-, Bi3-), for the first time, we design a promising antiperovskite alloy Mg3NAs0.5Bi0.5 with the quasi-direct band gap of 1.402 eV. Finally, we make a comprehensive comparison between antiperovskite nitrides and conventional halide perovskites for pointing out the future direction for device applications.

cond-mat.mtrl-sci↗

arXiv2021

Argus: A Fully Transparent Incentive System for Anti-Piracy Campaigns (Extended Version)

Anti-piracy is fundamentally a procedure that relies on collecting data from the open anonymous population, so how to incentivize credible reporting is a question at the center of the problem. Industrial alliances and companies are running anti-piracy incentive campaigns, but their effectiveness is publicly questioned due to the lack of transparency. We believe that full transparency of a campaign is necessary to truly incentivize people. It means that every role, e.g., content owner, licensee of the content, or every person in the open population, can understand the mechanism and be assured about its execution without trusting any single role. We see this as a distributed system problem. In this paper, we present Argus, a fully transparent incentive system for anti-piracy campaigns. The groundwork of Argus is to formulate the objectives for fully transparent incentive mechanisms, which securely and comprehensively consolidate the different interests of all roles. These objectives form the core of the Argus design, highlighted by our innovations about a Sybil-proof incentive function, a commit-and-reveal scheme, and an oblivious transfer scheme. In the implementation, we overcome a set of unavoidable obstacles to ensure security despite full transparency. Moreover, we effectively optimize several cryptographic operations so that the cost for a piracy reporting is reduced to an equivalent cost of sending about 14 ETH-transfer transactions to run on the public Ethereum network, which would otherwise correspond to thousands of transactions. With the security and practicality of Argus, we hope real-world anti-piracy campaigns will be truly effective by shifting to a fully transparent incentive mechanism.

cs.CR↗

arXiv2020

OpEvo: An Evolutionary Method for Tensor Operator Optimization

Training and inference efficiency of deep neural networks highly rely on the performance of tensor operators on hardware platforms. Manually optimizing tensor operators has limitations in terms of supporting new operators or hardware platforms. Therefore, automatically optimizing device code configurations of tensor operators is getting increasingly attractive. However, current methods for tensor operator optimization usually suffer from poor sample-efficiency due to the combinatorial search space. In this work, we propose a novel evolutionary method, OpEvo, which efficiently explores the search spaces of tensor operators by introducing a topology-aware mutation operation based on q-random walk to leverage the topological structures over the search spaces. Our comprehensive experiment results show that compared with state-of-the-art (SOTA) methods OpEvo can find the best configuration with the lowest variance and least efforts in the number of trials and wall-clock time. All code of this work is available online.

cs.LG↗

arXiv2019

TextNAS: A Neural Architecture Search Space tailored for Text Representation

Learning text representation is crucial for text classification and other language related tasks. There are a diverse set of text representation networks in the literature, and how to find the optimal one is a non-trivial problem. Recently, the emerging Neural Architecture Search (NAS) techniques have demonstrated good potential to solve the problem. Nevertheless, most of the existing works of NAS focus on the search algorithms and pay little attention to the search space. In this paper, we argue that the search space is also an important human prior to the success of NAS in different applications. Thus, we propose a novel search space tailored for text representation. Through automatic search, the discovered network architecture outperforms state-of-the-art models on various public datasets on text classification and natural language inference tasks. Furthermore, some of the design principles found in the automatic network agree well with human intuition.

cs.LG↗

arXiv2019

Time-Series Anomaly Detection Service at Microsoft

Large companies need to monitor various metrics (for example, Page Views and Revenue) of their applications and services in real time. At Microsoft, we develop a time-series anomaly detection service which helps customers to monitor the time-series continuously and alert for potential incidents on time. In this paper, we introduce the pipeline and algorithm of our anomaly detection service, which is designed to be accurate, efficient and general. The pipeline consists of three major modules, including data ingestion, experimentation platform and online compute. To tackle the problem of time-series anomaly detection, we propose a novel algorithm based on Spectral Residual (SR) and Convolutional Neural Network (CNN). Our work is the first attempt to borrow the SR model from visual saliency detection domain to time-series anomaly detection. Moreover, we innovatively combine SR and CNN together to improve the performance of SR model. Our approach achieves superior experimental results compared with state-of-the-art baselines on both public datasets and Microsoft production data.

cs.LG↗

arXiv2018

Defect-Fluorite Gd2Zr2O7 Ceramics under Helium Irradiation: Amorphization, Cell Volume Expansion, and Multi-stage Bubble Formation

Here, we report a study on the radiation resistance enhancement of Gd2Zr2O7 nanograin ceramics, in which amorphization, cell volume expansion and multi-stage helium (He) bubble formation are investigated and discussed. Gd2Zr2O7 ceramics with a series of grain sizes (55-221 nm) were synthesized and irradiated by 190 keV He ion beam up to a fluence of 5x10^17 ions/cm2. Both the degree of post irradiation cell volume expansion and the amorphization fraction appear to be size dependent. As the average grain size evolves from 55 to 221 nm, the degree of post irradiation cell volume expansion increases from 0.56 to 1.02 %, and the amorphization fraction increases from 6.8 to 11.1 %. Additionally, the threshold He concentrations (at. %) of bubbles at different formation stages and locations, including (1) bubbles at grain boundary, (2) bubble-chains and (3) ribbon-like bubbles within the grain, are all found to be much higher in the nanograin ceramic (55 nm) compared with that of the submicron sample (221 nm). We conclude that grain boundary plays a critical role in minimizing the structural defects, and inhibiting the multi-stage He bubble formation process.

cond-mat.mtrl-sci↗

arXiv2018

Survey and Performance Evaluation of the Upcoming Next Generation WLAN Standard - IEEE 802.11ax

With the ever-increasing demand for wireless traffic and quality of serives (QoS), wireless local area networks (WLANs) have developed into one of the most dominant wireless networks that fully influence human life. As the most widely used WLANs standard, Institute of Electrical and Electronics Engineers (IEEE) 802.11 will release the upcoming next generation WLANs standard amendment: IEEE 802.11ax. This article comprehensively surveys and analyzes the application scenarios, technical requirements, standardization process, key technologies, and performance evaluations of IEEE 802.11ax. Starting from the technical objectives and requirements of IEEE 802.11ax, this article pays special attention to high-dense deployment scenarios. After that, the key technologies of IEEE 802.11ax, including the physical layer (PHY) enhancements, multi-user (MU) medium access control (MU-MAC), spatial reuse (SR), and power efficiency are discussed in detail, covering both standardization technologies as well as the latest academic studies. Furthermore, performance requirements of IEEE 802.11ax are evaluated via a newly proposed systems and link-level integrated simulation platform (SLISP). Simulations results confirm that IEEE 802.11ax significantly improves the user experience in high-density deployment, while successfully achieves the average per user throughput requirement in project authorization request (PAR) by four times compared to the legacy IEEE 802.11. Finally, potential advancement beyond IEEE 802.11ax are discussed to complete this holistic study on the latest IEEE 802.11ax. To the best of our knowledge, this article is the first study to directly investigate and analyze the latest stable version of IEEE 802.11ax, and the first work to thoroughly and deeply evaluate the compliance of the performance requirements of IEEE 802.11ax.

cs.NI↗