arXiv subjects

Wei Lin

Publications and source records attributed to Wei Lin.

At least 289 records · Page 16Linked to original sources

arXiv2021

AdaBERT: Task-Adaptive BERT Compression with Differentiable Neural Architecture Search

Large pre-trained language models such as BERT have shown their effectiveness in various natural language processing tasks. However, the huge parameter size makes them difficult to be deployed in real-time applications that require quick inference with limited resources. Existing methods compress BERT into small models while such compression is task-independent, i.e., the same compressed BERT for all different downstream tasks. Motivated by the necessity and benefits of task-oriented BERT compression, we propose a novel compression method, AdaBERT, that leverages differentiable Neural Architecture Search to automatically compress BERT into task-adaptive small models for specific tasks. We incorporate a task-oriented knowledge distillation loss to provide search hints and an efficiency-aware loss as search constraints, which enables a good trade-off between efficiency and effectiveness for task-adaptive BERT compression. We evaluate AdaBERT on several NLP tasks, and the results demonstrate that those task-adaptive compressed models are 12.7x to 29.3x faster than BERT in inference time and 11.5x to 17.0x smaller in terms of parameter size, while comparable performance is maintained.

cs.CL↗

arXiv2020

INT8 Winograd Acceleration for Conv1D Equipped ASR Models Deployed on Mobile Devices

The intensive computation of Automatic Speech Recognition (ASR) models obstructs them from being deployed on mobile devices. In this paper, we present a novel quantized Winograd optimization pipeline, which combines the quantization and fast convolution to achieve efficient inference acceleration on mobile devices for ASR models. To avoid the information loss due to the combination of quantization and Winograd convolution, a Range-Scaled Quantization (RSQ) training method is proposed to expand the quantized numerical range and to distill knowledge from high-precision values. Moreover, an improved Conv1D equipped DFSMN (ConvDFSMN) model is designed for mobile deployment. We conduct extensive experiments on both ConvDFSMN and Wav2letter models. Results demonstrate the models can be effectively optimized with the proposed pipeline. Especially, Wav2letter achieves 1.48* speedup with an approximate 0.07% WER decrease on ARMv7-based mobile devices.

cs.SD↗

arXiv2020

Tag2Vec: Learning Tag Representations in Tag Networks

Network embedding is a method to learn low-dimensional representation vectors for nodes in complex networks. In real networks, nodes may have multiple tags but existing methods ignore the abundant semantic and hierarchical information of tags. This information is useful to many network applications and usually very stable. In this paper, we propose a tag representation learning model, Tag2Vec, which mixes nodes and tags into a hybrid network. Firstly, for tag networks, we define semantic distance as the proximity between tags and design a novel strategy, parameterized random walk, to generate context with semantic and hierarchical information of tags adaptively. Then, we propose hyperbolic Skip-gram model to express the complex hierarchical structure better with lower output dimensions. We evaluate our model on the NBER U.S. patent dataset and WordNet dataset. The results show that our model can learn tag representations with rich semantic information and it outperforms other baselines.

cs.SI↗

arXiv2020

One-shot Text Field Labeling using Attention and Belief Propagation for Structure Information Extraction

Structured information extraction from document images usually consists of three steps: text detection, text recognition, and text field labeling. While text detection and text recognition have been heavily studied and improved a lot in literature, text field labeling is less explored and still faces many challenges. Existing learning based methods for text labeling task usually require a large amount of labeled examples to train a specific model for each type of document. However, collecting large amounts of document images and labeling them is difficult and sometimes impossible due to privacy issues. Deploying separate models for each type of document also consumes a lot of resources. Facing these challenges, we explore one-shot learning for the text field labeling task. Existing one-shot learning methods for the task are mostly rule-based and have difficulty in labeling fields in crowded regions with few landmarks and fields consisting of multiple separate text regions. To alleviate these problems, we proposed a novel deep end-to-end trainable approach for one-shot text field labeling, which makes use of attention mechanism to transfer the layout information between document images. We further applied conditional random field on the transferred layout information for the refinement of field labeling. We collected and annotated a real-world one-shot field labeling dataset with a large variety of document types and conducted extensive experiments to examine the effectiveness of the proposed model. To stimulate research in this direction, the collected dataset and the one-shot model will be released1.

cs.CV↗

arXiv2020

Grasping Detection Network with Uncertainty Estimation for Confidence-Driven Semi-Supervised Domain Adaptation

Data-efficient domain adaptation with only a few labelled data is desired for many robotic applications, e.g., in grasping detection, the inference skill learned from a grasping dataset is not universal enough to directly apply on various other daily/industrial applications. This paper presents an approach enabling the easy domain adaptation through a novel grasping detection network with confidence-driven semi-supervised learning, where these two components deeply interact with each other. The proposed grasping detection network specially provides a prediction uncertainty estimation mechanism by leveraging on Feature Pyramid Network (FPN), and the mean-teacher semi-supervised learning utilizes such uncertainty information to emphasizing the consistency loss only for those unlabelled data with high confidence, which we referred it as the confidence-driven mean teacher. This approach largely prevents the student model to learn the incorrect/harmful information from the consistency loss, which speeds up the learning progress and improves the model accuracy. Our results show that the proposed network can achieve high success rate on the Cornell grasping dataset, and for domain adaptation with very limited data, the confidence-driven mean teacher outperforms the original mean teacher and direct training by more than 10% in evaluation loss especially for avoiding the overfitting and model diverging.

cs.CV↗

arXiv2020

NWPU-Crowd: A Large-Scale Benchmark for Crowd Counting and Localization

In the last decade, crowd counting and localization attract much attention of researchers due to its wide-spread applications, including crowd monitoring, public safety, space design, etc. Many Convolutional Neural Networks (CNN) are designed for tackling this task. However, currently released datasets are so small-scale that they can not meet the needs of the supervised CNN-based algorithms. To remedy this problem, we construct a large-scale congested crowd counting and localization dataset, NWPU-Crowd, consisting of 5,109 images, in a total of 2,133,375 annotated heads with points and boxes. Compared with other real-world datasets, it contains various illumination scenes and has the largest density range (0~20,033). Besides, a benchmark website is developed for impartially evaluating the different methods, which allows researchers to submit the results of the test set. Based on the proposed dataset, we further describe the data characteristics, evaluate the performance of some mainstream state-of-the-art (SOTA) methods, and analyze the new problems that arise on the new data. What's more, the benchmark is deployed at \url{https://www.crowdbenchmark.com/}, and the dataset/code/models/results are available at \url{https://gjy3035.github.io/NWPU-Crowd-Sample-Code/}.

cs.CV↗

arXiv2020

Pixel-wise Crowd Understanding via Synthetic Data

Crowd analysis via computer vision techniques is an important topic in the field of video surveillance, which has wide-spread applications including crowd monitoring, public safety, space design and so on. Pixel-wise crowd understanding is the most fundamental task in crowd analysis because of its finer results for video sequences or still images than other analysis tasks. Unfortunately, pixel-level understanding needs a large amount of labeled training data. Annotating them is an expensive work, which causes that current crowd datasets are small. As a result, most algorithms suffer from over-fitting to varying degrees. In this paper, take crowd counting and segmentation as examples from the pixel-wise crowd understanding, we attempt to remedy these problems from two aspects, namely data and methodology. Firstly, we develop a free data collector and labeler to generate synthetic and labeled crowd scenes in a computer game, Grand Theft Auto V. Then we use it to construct a large-scale, diverse synthetic crowd dataset, which is named as "GCC Dataset". Secondly, we propose two simple methods to improve the performance of crowd understanding via exploiting the synthetic data. To be specific, 1) supervised crowd understanding: pre-train a crowd analysis model on the synthetic data, then fine-tune it using the real data and labels, which makes the model perform better on the real world; 2) crowd understanding via domain adaptation: translate the synthetic data to photo-realistic images, then train the model on translated data and labels. As a result, the trained model works well in real crowd scenes.

cs.CV↗

arXiv2020

Auto-MAP: A DQN Framework for Exploring Distributed Execution Plans for DNN Workloads

The last decade has witnessed growth in the computational requirements for training deep neural networks. Current approaches (e.g., data/model parallelism, pipeline parallelism) parallelize training tasks onto multiple devices. However, these approaches always rely on specific deep learning frameworks and requires elaborate manual design, which make it difficult to maintain and share between different type of models. In this paper, we propose Auto-MAP, a framework for exploring distributed execution plans for DNN workloads, which can automatically discovering fast parallelization strategies through reinforcement learning on IR level of deep learning models. Efficient exploration remains a major challenge for reinforcement learning. We leverage DQN with task-specific pruning strategies to help efficiently explore the search space including optimized strategies. Our evaluation shows that Auto-MAP can find the optimal solution in two hours, while achieving better throughput on several NLP and convolution models.

cs.DC↗

arXiv2020

Graph Structural-topic Neural Network

Graph Convolutional Networks (GCNs) achieved tremendous success by effectively gathering local features for nodes. However, commonly do GCNs focus more on node features but less on graph structures within the neighborhood, especially higher-order structural patterns. However, such local structural patterns are shown to be indicative of node properties in numerous fields. In addition, it is not just single patterns, but the distribution over all these patterns matter, because networks are complex and the neighborhood of each node consists of a mixture of various nodes and structural patterns. Correspondingly, in this paper, we propose Graph Structural-topic Neural Network, abbreviated GraphSTONE, a GCN model that utilizes topic models of graphs, such that the structural topics capture indicative graph structures broadly from a probabilistic aspect rather than merely a few structures. Specifically, we build topic models upon graphs using anonymous walks and Graph Anchor LDA, an LDA variant that selects significant structural patterns first, so as to alleviate the complexity and generate structural topics efficiently. In addition, we design multi-view GCNs to unify node features and structural topic features and utilize structural topics to guide the aggregation. We evaluate our model through both quantitative and qualitative experiments, where our model exhibits promising performance, high efficiency, and clear interpretability.

cs.LG↗

arXiv2020

DAPPLE: A Pipelined Data Parallel Approach for Training Large Models

It is a challenging task to train large DNN models on sophisticated GPU platforms with diversified interconnect capabilities. Recently, pipelined training has been proposed as an effective approach for improving device utilization. However, there are still several tricky issues to address: improving computing efficiency while ensuring convergence, and reducing memory usage without incurring additional computing costs. We propose DAPPLE, a synchronous training framework which combines data parallelism and pipeline parallelism for large DNN models. It features a novel parallelization strategy planner to solve the partition and placement problems, and explores the optimal hybrid strategy of data and pipeline parallelism. We also propose a new runtime scheduling algorithm to reduce device memory usage, which is orthogonal to re-computation approach and does not come at the expense of training throughput. Experiments show that DAPPLE planner consistently outperforms strategies generated by PipeDream's planner by up to 3.23x under synchronous training scenarios, and DAPPLE runtime outperforms GPipe by 1.6x speedup of training throughput and reduces the memory consumption of 12% at the same time.

cs.DC↗

arXiv2020

Impact of intra and inter-cluster coupling balance on the performance of nonlinear networked systems

The dynamical and structural aspects of cluster synchronization (CS) in complex systems have been intensively investigated in recent years. Here, we study CS of dynamical systems with intra and inter-cluster couplings. We propose new metrics that describe the performance of such systems and evaluate them as a function of the strength of the couplings within and between clusters. We obtain analytical results that indicate that spectral differences between the Laplacian matrices associated with the partition between intra and inter-couplings directly affect the proposed metrics of system performance. Our results show that the dynamics of the system might exhibit an optimal balance that optimizes its performance. Our work provides new insights into the way specific symmetry properties relate to collective behavior, and could lead to new forms to increase the controllability of complex systems and to optimize their stability.

physics.soc-ph↗

arXiv2020

SwapText: Image Based Texts Transfer in Scenes

Swapping text in scene images while preserving original fonts, colors, sizes and background textures is a challenging task due to the complex interplay between different factors. In this work, we present SwapText, a three-stage framework to transfer texts across scene images. First, a novel text swapping network is proposed to replace text labels only in the foreground image. Second, a background completion network is learned to reconstruct background images. Finally, the generated foreground image and background image are used to generate the word image by the fusion network. Using the proposing framework, we can manipulate the texts of the input images even with severe geometric distortion. Qualitative and quantitative results are presented on several scene text datasets, including regular and irregular text datasets. We conducted extensive experiments to prove the usefulness of our method such as image based text translation, text image synthesis, etc.

cs.CV↗

arXiv2020

RPM-Oriented Query Rewriting Framework for E-commerce Keyword-Based Sponsored Search

Sponsored search optimizes revenue and relevance, which is estimated by Revenue Per Mille (RPM). Existing sponsored search models are all based on traditional statistical models, which have poor RPM performance when queries follow a heavy-tailed distribution. Here, we propose an RPM-oriented Query Rewriting Framework (RQRF) which outputs related bid keywords that can yield high RPM. RQRF embeds both queries and bid keywords to vectors in the same implicit space, converting the rewriting probability between each query and keyword to the distance between the two vectors. For label construction, we propose an RPM-oriented sample construction method, labeling keywords based on whether or not they can lead to high RPM. Extensive experiments are conducted to evaluate performance of RQRF. In a one month large-scale real-world traffic of e-commerce sponsored search system, the proposed model significantly outperforms traditional baseline.

cs.CL↗

arXiv2019

FusionStitching: Boosting Execution Efficiency of Memory Intensive Computations for DL Workloads

Performance optimization is the art of continuous seeking a harmonious mapping between the application domain and hardware. Recent years have witnessed a surge of deep learning (DL) applications in industry. Conventional wisdom for optimizing such workloads mainly focus on compute intensive ops (GEMM, Convolution, etc). Yet we show in this work, that the performance of memory intensive computations is vital to E2E performance in practical DL models. We propose \emph{FusionStitching}, a optimization framework capable of fusing memory intensive \emph{elementwise}, \emph{reduction} and fine grained \emph{GEMM/Batched-GEMM} ops, with or without data dependences, into large computation units, then mapping and transforming them into efficient GPU kernels. We formulate the fusion plan optimization as an integer linear programming (ILP) problem, and propose a set of empirical heuristics to reduce the combinatorial search space. In order to map optimized fusion plans to hardware, we propose a technique to effectively compose various groups of computations into a single GPU kernel, by fully leveraging on chip resources like scratchpads or registers. Experimental results on six benchmarks and four industry scale practical models are encouraging. Overall, \emph{FusionStitching} can reach up to 5.7x speedup compared to Tensorflow baseline, and achieves 1.25x to 1.85x performance speedups compared to current state of the art, with 1.4x on average (geometric mean).

cs.DC↗

arXiv2019

High-efficiency and high-power single-frequency fiber laser at 1.6 um based on cascaded energy-transfer pumping

In this paper, a technique combing cascaded energy-transfer pumping (CEP) method and master-oscillator power-amplifier (MOPA) configuration is proposed for power scaling of 1.6-um-band single-frequency fiber lasers (SFFLs), where the Er3+ ion has a limited gain. The CEP technique is fulfilled by coupling a primary signal light at 1.6 um and a C-band auxiliary laser. The numerical model of the fiber amplifier with the CEP technique reveals that the energy transfer process involves the pump competition and the in-band particle transition between the signal and auxiliary lights. Moreover, for the signal emission, the population density in the upper level is enhanced and the effective population inversion is achieved due to the CEP. A single-frequency MOPA laser at 1603 nm with an output power of 52.6 W is obtained experimentally. Besides, a slope efficiency of 30.4% is improved by more than 10% through the CEP technique. Both the output power and slope efficiency are by far the highest for 1.6-um-band SFFLs. Meanwhile, a laser linewidth of 5.2 kHz and a polarization-extinction ratio of ~18 dB are obtained at the maximum output power. The proposed technique provides an optional method of increasing the slope efficiency and power scaling for fiber lasers operating at L-band.

physics.optics↗

arXiv2019

Characterizing Deep Learning Training Workloads on Alibaba-PAI

Modern deep learning models have been exploited in various domains, including computer vision (CV), natural language processing (NLP), search and recommendation. In practical AI clusters, workloads training these models are run using software frameworks such as TensorFlow, Caffe, PyTorch and CNTK. One critical issue for efficiently operating practical AI clouds, is to characterize the computing and data transfer demands of these workloads, and more importantly, the training performance given the underlying software framework and hardware configurations. In this paper, we characterize deep learning training workloads from Platform of Artificial Intelligence (PAI) in Alibaba. We establish an analytical framework to investigate detailed execution time breakdown of various workloads using different training architectures, to identify performance bottleneck. Results show that weight/gradient communication during training takes almost 62% of the total execution time among all our workloads on average. The computation part, involving both GPU computing and memory access, are not the biggest bottleneck based on collective behavior of the workloads. We further evaluate attainable performance of the workloads on various potential software/hardware mappings, and explore implications on software architecture selection and hardware configurations. We identify that 60% of PS/Worker workloads can be potentially sped up when ported to the AllReduce architecture exploiting the high-speed NVLink for GPU interconnect, and on average 1.7X speedup can be achieved when Ethernet bandwidth is upgraded from 25 Gbps to 100 Gbps.

cs.PF↗

arXiv2019

DL2: A Deep Learning-driven Scheduler for Deep Learning Clusters

More and more companies have deployed machine learning (ML) clusters, where deep learning (DL) models are trained for providing various AI-driven services. Efficient resource scheduling is essential for maximal utilization of expensive DL clusters. Existing cluster schedulers either are agnostic to ML workload characteristics, or use scheduling heuristics based on operators' understanding of particular ML framework and workload, which are less efficient or not general enough. In this paper, we show that DL techniques can be adopted to design a generic and efficient scheduler. DL2 is a DL-driven scheduler for DL clusters, targeting global training job expedition by dynamically resizing resources allocated to jobs. DL2 advocates a joint supervised learning and reinforcement learning approach: a neural network is warmed up via offline supervised learning based on job traces produced by the existing cluster scheduler; then the neural network is plugged into the live DL cluster, fine-tuned by reinforcement learning carried out throughout the training progress of the DL jobs, and used for deciding job resource allocation in an online fashion. By applying past decisions made by the existing cluster scheduler in the preparatory supervised learning phase, our approach enables a smooth transition from existing scheduler, and renders a high-quality scheduler in minimizing average training completion time. We implement DL2 on Kubernetes and enable dynamic resource scaling in DL jobs on MXNet. Extensive evaluation shows that DL2 outperforms fairness scheduler (i.e., DRF) by 44.1% and expert heuristic scheduler (i.e., Optimus) by 17.5% in terms of average job completion time.

cs.LG↗

arXiv2019

GRID: a Student Project to Monitor the Transient Gamma-Ray Sky in the Multi-Messenger Astronomy Era

The Gamma-Ray Integrated Detectors (GRID) is a space mission concept dedicated to monitoring the transient gamma-ray sky in the energy range from 10 keV to 2 MeV using scintillation detectors onboard CubeSats in low Earth orbits. The primary targets of GRID are the gamma-ray bursts (GRBs) in the local universe. The scientific goal of GRID is, in synergy with ground-based gravitational wave (GW) detectors such as LIGO and VIRGO, to accumulate a sample of GRBs associated with the merger of two compact stars and study jets and related physics of those objects. It also involves observing and studying other gamma-ray transients such as long GRBs, soft gamma-ray repeaters, terrestrial gamma-ray flashes, and solar flares. With multiple CubeSats in various orbits, GRID is unaffected by the Earth occultation and serves as a full-time and all-sky monitor. Assuming a horizon of 200 Mpc for ground-based GW detectors, we expect to see a few associated GW-GRB events per year. With about 10 CubeSats in operation, GRID is capable of localizing a faint GRB like 170817A with a 90% error radius of about 10 degrees, through triangulation and flux modulation. GRID is proposed and developed by students, with considerable contribution from undergraduate students, and will remain operated as a student project in the future. The current GRID collaboration involves more than 20 institutes and keeps growing. On August 29th, the first GRID detector onboard a CubeSat was launched into a Sun-synchronous orbit and is currently under test.

astro-ph.IM↗