arXiv subjects

Xiaodong Yu

Publications and source records attributed to Xiaodong Yu.

66 records · Page 4Linked to original sources

arXiv2022

Learning to Decompose: Hypothetical Question Decomposition Based on Comparable Texts

Explicit decomposition modeling, which involves breaking down complex tasks into more straightforward and often more interpretable sub-tasks, has long been a central theme in developing robust and interpretable NLU systems. However, despite the many datasets and resources built as part of this effort, the majority have small-scale annotations and limited scope, which is insufficient to solve general decomposition tasks. In this paper, we look at large-scale intermediate pre-training of decomposition-based transformers using distant supervision from comparable texts, particularly large-scale parallel news. We show that with such intermediate pre-training, developing robust decomposition-based models for a diverse range of tasks becomes more feasible. For example, on semantic parsing, our model, DecompT5, improves 20% to 30% on two datasets, Overnight and TORQUE, over the baseline language model. We further use DecompT5 to build a novel decomposition-based QA system named DecompEntail, improving over state-of-the-art models, including GPT-3, on both HotpotQA and StrategyQA by 8% and 4%, respectively.

cs.CL↗

arXiv2022

InsPose: Instance-Aware Networks for Single-Stage Multi-Person Pose Estimation

Multi-person pose estimation is an attractive and challenging task. Existing methods are mostly based on two-stage frameworks, which include top-down and bottom-up methods. Two-stage methods either suffer from high computational redundancy for additional person detectors or they need to group keypoints heuristically after predicting all the instance-agnostic keypoints. The single-stage paradigm aims to simplify the multi-person pose estimation pipeline and receives a lot of attention. However, recent single-stage methods have the limitation of low performance due to the difficulty of regressing various full-body poses from a single feature vector. Different from previous solutions that involve complex heuristic designs, we present a simple yet effective solution by employing instance-aware dynamic networks. Specifically, we propose an instance-aware module to adaptively adjust (part of) the network parameters for each instance. Our solution can significantly increase the capacity and adaptive-ability of the network for recognizing various poses, while maintaining a compact end-to-end trainable pipeline. Extensive experiments on the MS-COCO dataset demonstrate that our method achieves significant improvement over existing single-stage methods, and makes a better balance of accuracy and efficiency compared to the state-of-the-art two-stage approaches. The code and models are available at \url{https://github.com/hikvision-research/opera}.

cs.CV↗

arXiv2022

Optimizing Huffman Decoding for Error-Bounded Lossy Compression on GPUs

More and more HPC applications require fast and effective compression techniques to handle large volumes of data in storage and transmission. Not only do these applications need to compress the data effectively during simulation, but they also need to perform decompression efficiently for post hoc analysis. SZ is an error-bounded lossy compressor for scientific data, and cuSZ is a version of SZ designed to take advantage of the GPU's power. At present, cuSZ's compression performance has been optimized significantly while its decompression still suffers considerably lower performance because of its sophisticated lossless compression step -- a customized Huffman decoding. In this work, we aim to significantly improve the Huffman decoding performance for cuSZ, thus improving the overall decompression performance in turn. To this end, we first investigate two state-of-the-art GPU Huffman decoders in depth. Then, we propose a deep architectural optimization for both algorithms. Specifically, we take full advantage of CUDA GPU architectures by using shared memory on decoding/writing phases, online tuning the amount of shared memory to use, improving memory access patterns, and reducing warp divergence. Finally, we evaluate our optimized decoders on an Nvidia V100 GPU using eight representative scientific datasets. Our new decoding solution obtains an average speedup of 3.64X over cuSZ's Huffman decoder and improves its overall decompression performance by 2.43X on average.

cs.DC↗

arXiv2022

SZx: an Ultra-fast Error-bounded Lossy Compressor for Scientific Datasets

Today's scientific high performance computing (HPC) applications or advanced instruments are producing vast volumes of data across a wide range of domains, which introduces a serious burden on data transfer and storage. Error-bounded lossy compression has been developed and widely used in scientific community, because not only can it significantly reduce the data volumes but it can also strictly control the data distortion based on the use-specified error bound. Existing lossy compressors, however, cannot offer ultra-fast compression speed, which is highly demanded by quite a few applications or use-cases (such as in-memory compression and online instrument data compression). In this paper, we propose a novel ultra-fast error-bounded lossy compressor, which can obtain fairly high compression performance on both CPU and GPU, also with reasonably high compression ratios. The key contributions are three-fold: (1) We propose a novel, generic ultra-fast error-bounded lossy compression framework -- called UFZ, by confining our design to be composed of only super-lightweight operations such as bitwise and addition/subtraction operation, still keeping a certain high compression ratio. (2) We implement UFZ on both CPU and GPU and optimize the performance according to their architectures carefully. (3) We perform a comprehensive evaluation with 6 real-world production-level scientific datasets on both CPU and GPU. Experiments show that UFZ is 2~16X as fast as the second-fastest existing error-bounded lossy compressor (either SZ or ZFP) on CPU and GPU, with respect to both compression and decompression.

cs.DC↗

arXiv2021

SOIT: Segmenting Objects with Instance-Aware Transformers

This paper presents an end-to-end instance segmentation framework, termed SOIT, that Segments Objects with Instance-aware Transformers. Inspired by DETR \cite{carion2020end}, our method views instance segmentation as a direct set prediction problem and effectively removes the need for many hand-crafted components like RoI cropping, one-to-many label assignment, and non-maximum suppression (NMS). In SOIT, multiple queries are learned to directly reason a set of object embeddings of semantic category, bounding-box location, and pixel-wise mask in parallel under the global image context. The class and bounding-box can be easily embedded by a fixed-length vector. The pixel-wise mask, especially, is embedded by a group of parameters to construct a lightweight instance-aware transformer. Afterward, a full-resolution mask is produced by the instance-aware transformer without involving any RoI-based operation. Overall, SOIT introduces a simple single-stage instance segmentation framework that is both RoI- and NMS-free. Experimental results on the MS COCO dataset demonstrate that SOIT outperforms state-of-the-art instance segmentation approaches significantly. Moreover, the joint learning of multiple tasks in a unified query embedding can also substantially improve the detection performance. Code is available at \url{https://github.com/yuxiaodongHRI/SOIT}.

cs.CV↗

arXiv2021

High-Performance Ptychographic Reconstruction with Federated Facilities

Beamlines at synchrotron light source facilities are powerful scientific instruments used to image samples and observe phenomena at high spatial and temporal resolutions. Typically, these facilities are equipped only with modest compute resources for the analysis of generated experimental datasets. However, high data rate experiments can easily generate data in volumes that take days (or even weeks) to process on those local resources. To address this challenge, we present a system that unifies leadership computing and experimental facilities by enabling the automated establishment of data analysis pipelines that extend from edge data acquisition systems at synchrotron beamlines to remote computing facilities; under the covers, our system uses Globus Auth authentication to minimize user interaction, funcX to run user-defined functions on supercomputers, and Globus Flows to define and execute workflows. We describe the application of this system to ptychography, an ultra-high-resolution coherent diffraction imaging technique that can produce 100s of gigabytes to terabytes in a single experiment. When deployed on the DGX A100 ThetaGPU cluster at the Argonne Leadership Computing Facility and a microscopy beamline at the Advanced Photon Source, our system performs analysis as an experiment progresses to provide timely feedback.

cs.DC↗

arXiv2021

Optimizing Error-Bounded Lossy Compression for Scientific Data on GPUs

Error-bounded lossy compression is a critical technique for significantly reducing scientific data volumes. With ever-emerging heterogeneous high-performance computing (HPC) architecture, GPU-accelerated error-bounded compressors (such as cuSZ+ and cuZFP) have been developed. However, they suffer from either low performance or low compression ratios. To this end, we propose cuSZ+ to target both high compression ratios and throughputs. We identify that data sparsity and data smoothness are key factors for high compression throughputs. Our key contributions in this work are fourfold: (1) We propose an efficient compression workflow to adaptively perform run-length encoding and/or variable-length encoding. (2) We derive Lorenzo reconstruction in decompression as multidimensional partial-sum computation and propose a fine-grained Lorenzo reconstruction algorithm for GPU architectures. (3) We carefully optimize each of cuSZ+ kernels by leveraging state-of-the-art CUDA parallel primitives. (4) We evaluate cuSZ+ using seven real-world HPC application datasets on V100 and A100 GPUs. Experiments show cuSZ+ improves the compression throughputs and ratios by up to 18.4X and 5.3X, respectively, over cuSZ on the tested datasets.

cs.DC↗

arXiv2021

Scalable and accurate multi-GPU based image reconstruction of large-scale ptychography data

While the advances in synchrotron light sources, together with the development of focusing optics and detectors, allow nanoscale ptychographic imaging of materials and biological specimens, the corresponding experiments can yield terabyte-scale large volumes of data that can impose a heavy burden on the computing platform. While Graphical Processing Units (GPUs) provide high performance for such large-scale ptychography datasets, a single GPU is typically insufficient for analysis and reconstruction. Several existing works have considered leveraging multiple GPUs to accelerate the ptychographic reconstruction. However, they utilize only Message Passing Interface (MPI) to handle the communications between GPUs. It poses inefficiency for the configuration that has multiple GPUs in a single node, especially while processing a single large projection, since it provides no optimizations to handle the heterogeneous GPU interconnections containing both low-speed links, e.g., PCIe, and high-speed links, e.g., NVLink. In this paper, we provide a multi-GPU implementation that can effectively solve large-scale ptychographic reconstruction problem with optimized performance on intra-node multi-GPU. We focus on the conventional maximum-likelihood reconstruction problem using conjugate-gradient (CG) for the solution and propose a novel hybrid parallelization model to address the performance bottlenecks in CG solver. Accordingly, we develop a tool called PtyGer (Ptychographic GPU(multiple)-based reconstruction), implementing our hybrid parallelization model design. The comprehensive evaluation verifies that PtyGer can fully preserve the original algorithm's accuracy while achieving outstanding intra-node GPU scalability.

cs.DC↗

arXiv2020

Design Challenges in Low-resource Cross-lingual Entity Linking

Cross-lingual Entity Linking (XEL), the problem of grounding mentions of entities in a foreign language text into an English knowledge base such as Wikipedia, has seen a lot of research in recent years, with a range of promising techniques. However, current techniques do not rise to the challenges introduced by text in low-resource languages (LRL) and, surprisingly, fail to generalize to text not taken from Wikipedia, on which they are usually trained. This paper provides a thorough analysis of low-resource XEL techniques, focusing on the key step of identifying candidate English Wikipedia titles that correspond to a given foreign language mention. Our analysis indicates that current methods are limited by their reliance on Wikipedia's interlanguage links and thus suffer when the foreign language's Wikipedia is small. We conclude that the LRL setting requires the use of outside-Wikipedia cross-lingual resources and present a simple yet effective zero-shot XEL system, QuEL, that utilizes search engines query logs. With experiments on 25 languages, QuEL~shows an average increase of 25\% in gold candidate recall and of 13\% in end-to-end linking accuracy over state-of-the-art baselines.

cs.CL↗

arXiv2020

Combinatorial Laser Molecular Beam Epitaxy System Integrated with Specialized Low-temperature Scanning Tunneling Microscopy

We present a newly developed facility, comprised of a combinatorial laser molecular beam epitaxy system and an in-situ scanning tunneling microscopy (STM). This facility aims at accelerating the materials research in a highly efficient way, by advanced high-throughput film synthesis techniques and subsequent fast characterization of surface morphology and electronic states. Compared with uniform films deposited by conventional methods, the so-called combinatorial thin films will be beneficial to determining the accurate phase diagrams of different materials due to the improved control of parameters such as chemical substitution and sample thickness resulting from a rotarymask method. A specially designed STM working under low-temperature and ultra-high vacuum conditions is optimized for the characterization of combinatorial thin films, in an XY coarse motion range of 15 mm $\times$ 15 mm and with sub-micrometer location precision. The overall configuration as well as some key aspects like sample holder design, scanner head, and sample/tip/target transfer mechanism are described in detail. The performance of the device is demonstrated by synthesizing high-quality superconducting FeSe thin films with gradient thickness, imaging surfaces of highly oriented pyrolytic graphite, Au (111), Bi2Sr2CaCu2O8+δ (BSCCO) and FeSe. In addition, we have also obtained clean noise spectra of tunneling junctions and the superconducting energy gap of BSCCO. The successful manufacturing of such a facility opens a new window for the next generation of equipment designed for experimental materials research.

physics.app-ph↗

arXiv2018

On the Strength of Character Language Models for Multilingual Named Entity Recognition

Character-level patterns have been widely used as features in English Named Entity Recognition (NER) systems. However, to date there has been no direct investigation of the inherent differences between name and non-name tokens in text, nor whether this property holds across multiple languages. This paper analyzes the capabilities of corpus-agnostic Character-level Language Models (CLMs) in the binary task of distinguishing name tokens from non-name tokens. We demonstrate that CLMs provide a simple and powerful model for capturing these differences, identifying named entity tokens in a diverse set of languages at close to the performance of full NER systems. Moreover, by adding very simple CLM-based features we can significantly improve the performance of an off-the-shelf NER system for multiple languages.

cs.CL↗

arXiv2018

Structures and Physical Properties of CsV$_2$Se$_{2-x}$O and V$_2$Se$_2$O

By using solid-state reactions, we successfully synthesize new oxyselenides CsV$_2$Se$_{2-x}$O (x = 0, 0.5). These compounds containing V$_2$O planar layers with a square lattice crystallize in the CeCr$_2$Si$_2$C structure with the space group of $P4/mmm$. Another new compound V$_2$Se$_2$O which crystallizes in space group $I4/mmm$ is fabricated by topochemical deintercalation of cesium from CsV$_2$Se$_2$O powder with iodine in tetrahydrofuran(THF). Resistivity measurements show a semiconducting behavior for CsV$_2$Se$_2$O, while a metallic behavior for CsV$_2$Se$_{1.5}$O, and an insulating feature for V$_2$Se$_2$O. A charge- or spin-density wave-like anomaly has been observed at 168 K for CsV$_2$Se$_2$O and 150 K for CsV$_2$Se$_{1.5}$O, respectively. And these anomalies are also confirmed by the magnetic susceptibility measurements. The resistivity in V$_2$Se$_2$O exhibits an anomalous log(1/$T$) temperature dependence, which is similar to the case in parent phase or very underdoped cuprates indicating the involvement of strong correlation. Magnetic susceptibility measurements show that the magnetic moment per V-site in V$_2$Se$_2$O is much larger than that of CsV$_2$Se$_{2-x}$O, which again suggests the correlation induced localization effect in the former.

cond-mat.mtrl-sci↗