Zizhong Chen

Zizhong Chen, Jack Dongarra

Let $G_{m \times n}$ be an $m \times n$ real random matrix whose elements are independent and identically distributed standard normal random variables, and let $κ_2(G_{m \times n})$ be the 2-norm condition number of $G_{m \times n}$. We prove that, for any $m \geq 2$, $n \geq 2$ and $x \geq |n-m|+1$, $κ_2(G_{m \times n})$ satisfies $ \frac{1}{\sqrt{2π}} ({c}/{x})^{|n-m|+1} < P(\frac{κ_2(G_{m \times n})} {{n}/{(|n-m|+1)}}> x) < \frac{1}{\sqrt{2π}} ({C}/{x})^{|n-m|+1}, $ where $0.245 \leq c \leq 2.000$ and $ 5.013 \leq C \leq 6.414$ are universal positive constants independent of $m$, $n$ and $x$. Moreover, for any $m \geq 2$ and $n \geq 2$, $ E(\logκ_2(G_{m \times n})) < \log \frac{n}{|n-m|+1} + 2.258. $ A similar pair of results for complex Gaussian random matrices is also established.

Yujia Zhai, Chengquan Jiang, Leyuan Wang et al.

Transformers have become keystone models in natural language processing over the past decade. They have achieved great popularity in deep learning applications, but the increasing sizes of the parameter spaces required by transformer models generate a commensurate need to accelerate performance. Natural language processing problems are also routinely faced with variable-length sequences, as word counts commonly vary among sentences. Existing deep learning frameworks pad variable-length sequences to a maximal length, which adds significant memory and computational overhead. In this paper, we present ByteTransformer, a high-performance transformer boosted for variable-length inputs. We propose a padding-free algorithm that liberates the entire transformer from redundant computations on zero padded tokens. In addition to algorithmic-level optimization, we provide architecture-aware optimizations for transformer functional modules, especially the performance-critical algorithm Multi-Head Attention (MHA). Experimental results on an NVIDIA A100 GPU with variable-length sequence inputs validate that our fused MHA outperforms PyTorch by 6.13x. The end-to-end performance of ByteTransformer for a forward BERT transformer surpasses state-of-the-art transformer frameworks, such as PyTorch JIT, TensorFlow XLA, Tencent TurboTransformer, Microsoft DeepSpeed-Inference and NVIDIA FasterTransformer, by 87\%, 131\%, 138\%, 74\% and 55\%, respectively. We also demonstrate the general applicability of our optimization methods to other BERT-like models, including ALBERT, DistilBERT, and DeBERTa.

Jinyang Liu, Sheng Di, Sian Jin et al.

The fast growth of computational power and scales of modern super-computing systems have raised great challenges for the management of exascale scientific data. To maintain the usability of scientific data, error-bound lossy compression is proposed and developed as an essential technique for the size reduction of scientific data with constrained data distortion. Among the diverse datasets generated by various scientific simulations, certain datasets cannot be effectively compressed by existing error-bounded lossy compressors with traditional techniques. The recent success of Artificial Intelligence has inspired several researchers to integrate neural networks into error-bounded lossy compressors. However, those works still suffer from limited compression ratios and/or extremely low efficiencies. To address those issues and improve the compression on the hard-to-compress datasets, in this paper, we propose SRN-SZ, which is a deep learning-based scientific error-bounded lossy compressor leveraging the hierarchical data grid expansion paradigm implemented by super-resolution neural networks. SRN-SZ applies the most advanced super-resolution network HAT for its compression, which is free of time-costing per-data training. In experiments compared with various state-of-the-art compressors, SRN-SZ achieves up to 75% compression ratio improvements under the same error bound and up to 80% compression ratio improvements under the same PSNR than the second-best compressor.

Shixun Wu, Yitong Ding, Yujia Zhai et al.

K-means is a widely used algorithm in clustering, however, its efficiency is primarily constrained by the computational cost of distance computing. Existing implementations suffer from suboptimal utilization of computational units and lack resilience against soft errors. To address these challenges, we introduce FT K-means, a high-performance GPU-accelerated implementation of K-means with online fault tolerance. We first present a stepwise optimization strategy that achieves competitive performance compared to NVIDIA's cuML library. We further improve FT K-means with a template-based code generation framework that supports different data types and adapts to different input shapes. A novel warp-level tensor-core error correction scheme is proposed to address the failure of existing fault tolerance methods due to memory asynchronization during copy operations. Our experimental evaluations on NVIDIA T4 GPU and A100 GPU demonstrate that FT K-means without fault tolerance outperforms cuML's K-means implementation, showing a performance increase of 10\%-300\% in scenarios involving irregular data shapes. Moreover, the fault tolerance feature of FT K-means introduces only an overhead of 11\%, maintaining robust performance even with tens of errors injected per second.

Huangliang Dai, Shixun Wu, Jiajun Huang et al.

Transformer models rely on High-Performance Computing (HPC) resources for inference, where soft errors are inevitable in large-scale systems, making the reliability of the model particularly critical. Existing fault tolerance frameworks for Transformers are designed at the operation level without architectural optimization, leading to significant computational and memory overhead, which in turn reduces protection efficiency and limits scalability to larger models. In this paper, we implement module-level protection for Transformers by treating the operations within the attention module as a single kernel and applying end-to-end fault tolerance. This method provides unified protection across multi-step computations, while achieving comprehensive coverage of potential errors in the nonlinear computations. For linear modules, we design a strided algorithm-based fault tolerance (ABFT) that avoids inter-thread communication. Experimental results show that our end-to-end fault tolerance achieves up to 7.56x speedup over traditional methods with an average fault tolerance overhead of 13.9%.

Dawei Dai, Xiaoyu Tang, Shuyin Xia et al.

Fine-grained sketch-based image retrieval (FG-SBIR) addresses the problem of retrieving a particular photo in a given query sketch. However, its widespread applicability is limited by the fact that it is difficult to draw a complete sketch for most people, and the drawing process often takes time. In this study, we aim to retrieve the target photo with the least number of strokes possible (incomplete sketch), named on-the-fly FG-SBIR (Bhunia et al. 2020), which starts retrieving at each stroke as soon as the drawing begins. We consider that there is a significant correlation among these incomplete sketches in the sketch drawing episode of each photo. To learn more efficient joint embedding space shared between the photo and its incomplete sketches, we propose a multi-granularity association learning framework that further optimizes the embedding space of all incomplete sketches. Specifically, based on the integrity of the sketch, we can divide a complete sketch episode into several stages, each of which corresponds to a simple linear mapping layer. Moreover, our framework guides the vector space representation of the current sketch to approximate that of its later sketches to realize the retrieval performance of the sketch with fewer strokes to approach that of the sketch with more strokes. In the experiments, we proposed more realistic challenges, and our method achieved superior early retrieval efficiency over the state-of-the-art methods and alternative baselines on two publicly available fine-grained sketch retrieval datasets.

Shuyin Xia, Cheng Wang, Guoyin Wang et al.

Pawlak rough set and neighborhood rough set are the two most common rough set theoretical models. Pawlak can use equivalence classes to represent knowledge, but it cannot process continuous data; neighborhood rough sets can process continuous data, but it loses the ability of using equivalence classes to represent knowledge. To this end, this paper presents a granular-ball rough set based on the granular-ball computing. The granular-ball rough set can simultaneously represent Pawlak rough sets, and the neighborhood rough set, so as to realize the unified representation of the two. This makes the granular-ball rough set not only can deal with continuous data, but also can use equivalence classes for knowledge representation. In addition, we propose an implementation algorithms of granular-ball rough sets. The experimental results on benchmark datasets demonstrate that, due to the combination of the robustness and adaptability of the granular-ball computing, the learning accuracy of the granular-ball rough set has been greatly improved compared with the Pawlak rough set and the traditional neighborhood rough set. The granular-ball rough set also outperforms nine popular or the state-of-the-art feature selection methods.

Shuyin Xia, Xinyu Bai, Guoyin Wang et al.

This paper present a strong data mining method based on rough set, which can realize feature selection, classification and knowledge representation at the same time. Rough set has good interpretability, and is a popular method for feature selections. But low efficiency and low accuracy are its main drawbacks that limits its application ability. In this paper,corresponding to the accuracy, we first find the ineffectiveness of rough set because of overfitting, especially in processing noise attribute, and propose a robust measurement for an attribute, called relative importance.we proposed the concept of "rough concept tree" for knowledge representation and classification. Experimental results on public benchmark data sets show that the proposed framework achieves higher accurcy than seven popular or the state-of-the-art feature selection methods.

Yujia Zhai, Mohannad Ibrahim, Yiqin Qiu et al.

Homomorphic Encryption (HE) is an emerging encryption scheme that allows computations to be performed directly on encrypted messages. This property provides promising applications such as privacy-preserving deep learning and cloud computing. Prior works have been proposed to enable practical privacy-preserving applications with architectural-aware optimizations on CPUs, GPUs and FPGAs. However, there is no systematic optimization for the whole HE pipeline on Intel GPUs. In this paper, we present the first-ever SYCL-based GPU backend for Microsoft SEAL APIs. We perform optimizations from instruction level, algorithmic level and application level to accelerate our HE library based on the Cheon, Kim, Kimand Song (CKKS) scheme on Intel GPUs. The performance is validated on two latest Intel GPUs. Experimental results show that our staged optimizations together with optimizations including low-level optimizations and kernel fusion accelerate the Number Theoretic Transform (NTT), a key algorithm for HE, by up to 9.93X compared with the naïve GPU baseline. The roofline analysis confirms that our optimized NTT reaches 79.8% and85.7% of the peak performance on two GPU devices. Through the highly optimized NTT and the assembly-level optimization, we obtain 2.32X - 3.05X acceleration for HE evaluation routines. In addition, our all-together systematic optimizations improve the performance of encrypted element-wise polynomial matrix multiplication application by up to 3.10X.

Jinyang Liu, Sheng Di, Kai Zhao et al.

Error-bounded lossy compression is becoming an indispensable technique for the success of today's scientific projects with vast volumes of data produced during simulations or instrument data acquisitions. Not only can it significantly reduce data size, but it also can control the compression errors based on user-specified error bounds. Autoencoder (AE) models have been widely used in image compression, but few AE-based compression approaches support error-bounding features, which are highly required by scientific applications. To address this issue, we explore using convolutional autoencoders to improve error-bounded lossy compression for scientific data, with the following three key contributions. (1) We provide an in-depth investigation of the characteristics of various autoencoder models and develop an error-bounded autoencoder-based framework in terms of the SZ model. (2) We optimize the compression quality for the main stages in our designed AE-based error-bounded compression framework, fine-tuning the block sizes and latent sizes and also optimizing the compression efficiency of latent vectors. (3) We evaluate our proposed solution using five real-world scientific datasets and compare them with six other related works. Experiments show that our solution exhibits a very competitive compression quality among all the compressors in our tests. In absolute terms, it can obtain a much better compression quality (100% ~ 800% improvement in compression ratio with the same data distortion) compared with SZ2.1 and ZFP in cases with a high compression ratio.

Shuyin Xia, Daowan Peng, Deyu Meng et al.

This paper presents a novel accelerated exact k-means algorithm called the Ball k-means algorithm, which uses a ball to describe a cluster, focusing on reducing the point-centroid distance computation. The Ball k-means can accurately find the neighbor clusters for each cluster resulting distance computations only between a point and its neighbor clusters' centroids instead of all centroids. Moreover, each cluster can be divided into a stable area and an active area, and the later one can be further divided into annulus areas. The assigned cluster of the points in the stable area is not changed in the current iteration while the points in the annulus area will be adjusted within a few neighbor clusters in the current iteration. Also, there are no upper or lower bounds in the proposed Ball k-means. Furthermore, reducing centroid-centroid distance computation between iterations makes it efficient for large k clustering. The fast speed, no extra parameters and simple design of the Ball k-means make it an all-around replacement of the naive k-means algorithm.

Kai Zhao, Sheng Di, Sihuan Li et al.

Convolutional neural networks (CNNs) are becoming more and more important for solving challenging and critical problems in many fields. CNN inference applications have been deployed in safety-critical systems, which may suffer from soft errors caused by high-energy particles, high temperature, or abnormal voltage. Of critical importance is ensuring the stability of the CNN inference process against soft errors. Traditional fault tolerance methods are not suitable for CNN inference because error-correcting code is unable to protect computational components, instruction duplication techniques incur high overhead, and existing algorithm-based fault tolerance (ABFT) techniques cannot protect all convolution implementations. In this paper, we focus on how to protect the CNN inference process against soft errors as efficiently as possible, with the following three contributions. (1) We propose several systematic ABFT schemes based on checksum techniques and analyze their fault protection ability and runtime thoroughly.Unlike traditional ABFT based on matrix-matrix multiplication, our schemes support any convolution implementations. (2) We design a novel workflow integrating all the proposed schemes to obtain a high detection/correction ability with limited total runtime overhead. (3) We perform our evaluation using ImageNet with well-known CNN models including AlexNet, VGG-19, ResNet-18, and YOLOv2. Experimental results demonstrate that our implementation can handle soft errors with very limited runtime overhead (4%~8% in both error-free and error-injected situations).

Jinyang Liu, Yujia Zhai, Zizhong Chen

Neural Network based models have been state-of-the-art models for various Natural Language Processing tasks, however, the input and output dimension problem in the networks has still not been fully resolved, especially in text generation tasks (e.g. Machine Translation, Text Summarization), in which input and output both have huge sizes of vocabularies. Therefore, input-output embedding weight sharing has been introduced and adopted widely, which remains to be improved. Based on linear algebra and statistical theories, this paper locates the shortcoming of existed input-output embedding weight sharing method, then raises methods for improving input-output weight shared embedding, among which methods of normalization of embedding weight matrices show best performance. These methods are nearly computational cost-free, can get combined with other embedding techniques, and show good effectiveness when applied on state-of-the-art Neural Network models. For Transformer-big models, the normalization techniques can get at best 0.6 BLEU improvement compared to the original version of model on WMT'16 En-De dataset, and similar BLEU improvements on IWSLT 14' datasets. For DynamicConv models, 0.5 BLEU improvement can be attained on WMT'16 En-De dataset, and 0.41 BLEU improvement on IWSLT 14' De-En translation task is achieved.