Search

Your search keyword '"Zhu, Xueying"' showing total 191 results
Start Over Author "Zhu, Xueying" Publication Year Range Last 10 years
191 results on '"Zhu, Xueying"'

3. P4SGD: Programmable Switch Enhanced Model-Parallel Training on Generalized Linear Models on Distributed FPGAs

Author

Huang, Hongjing, Li, Yingtao, Sun, Jie, Zhu, Xueying, Zhang, Jie, Luo, Liang, Li, Jialin, and Wang, Zeke

Subjects
Computer Science - Distributed, Parallel, and Cluster Computing
Abstract

Generalized linear models (GLMs) are a widely utilized family of machine learning models in real-world applications. As data size increases, it is essential to perform efficient distributed training for these models. However, existing systems for distributed training have a high cost for communication and often use large batch sizes to balance computation and communication, which negatively affects convergence. Therefore, we argue for an efficient distributed GLM training system that strives to achieve linear scalability, while keeping batch size reasonably low. As a start, we propose P4SGD, a distributed heterogeneous training system that efficiently trains GLMs through model parallelism between distributed FPGAs and through forward-communication-backward pipeline parallelism within an FPGA. Moreover, we propose a light-weight, latency-centric in-switch aggregation protocol to minimize the latency of the AllReduce operation between distributed FPGAs, powered by a programmable switch. As such, to our knowledge, P4SGD is the first solution that achieves almost linear scalability between distributed accelerators through model parallelism. We implement P4SGD on eight Xilinx U280 FPGAs and a Tofino P4 switch. Our experiments show P4SGD converges up to 6.5X faster than the state-of-the-art GPU counterpar.

Published
2023

17. A cerium-doped tungsten trioxide-functionalized sensing platform for photoelectrochemical detection of ascorbic acid with high sensitivity.

Author

Zhu, Xueying, Liang, Tikai, and Tang, Dianping

Subjects
*VITAMIN C, *TUNGSTEN trioxide, *ELECTRON traps, *ELECTRON donors, *TRANSMISSION electron microscopy
Abstract

A highly efficient photoelectrochemical (PEC) strategy was proposed for the determination of ascorbic acid (AA). Cerium-doped tungsten trioxide (Ce-WO3) microrods were synthesized by a hydrothermal method and further characterized through transmission electron microscopy (TEM), X-ray diffraction (XRD), and Raman spectroscopy. Thereafter, they were deposited onto a cleaned fluorine-doped tin oxide (FTO) glass forming the working electrode as the photoactive material. Under strong visible light irradiation, the resulting PEC sensing platform generated the corresponding electron–hole pairs, converting light signals into electrical signals. Ascorbic acid served as a good electron donor to trap holes for improvement of photocurrent responses on Ce-WO3/FTO. Besides, the strength of photocurrent signals versus the logarithm of ascorbic acid concentration showed a good linearity over the ascorbic acid concentration range of 100–4000 nM and the limit of detection (LOD) was estimated to be 28.6 nM. Importantly, this PEC sensor had a fast response, high sensitivity, and distinguished selectivity for detecting ascorbic acid. In addition, it also had the features of being simple to fabricate, low production cost, and portable, which made it a promising means of ascorbic acid determination. [ABSTRACT FROM AUTHOR]

Published
2024
Full Text
View/download PDF

21. A model for the noninvasive, habitat-inclusive estimation of upper limit abundance for synanthropes, exemplified by M. fascicularis

Author

Koch Liston, André L., primary, Zhu, Xueying, additional, Bang, Tran V., additional, Phiapalath, Phaivanh, additional, Hun, Seiha, additional, Ahmed, Tanvir, additional, Hasan, Sabit, additional, Biswas, Sajib, additional, Nath, Shimul, additional, Ahmed, Toufique, additional, Ilham, Kurnia, additional, Lwin, Ngwe, additional, Frechette, Jackson L., additional, Hon, Naven, additional, Agger, Cain, additional, Ai, Suzuki, additional, Auda, Emeline, additional, Gazagne, Eva, additional, Kamler, Jan F., additional, Groenenberg, Milou, additional, Banet-Eugene, Sarah, additional, Challis, Neil, additional, Vibol, Neth, additional, Leroux, Nicole, additional, Sinovas, Pablo, additional, Reaksmey, Sophatt, additional, Muñoz, Vanessa H., additional, Lappan, Susan, additional, Zainol, Zaki, additional, Albanese, Valeria, additional, Alexiadou, Athanasia, additional, Nielsen, Daniel R. K., additional, Holzner, Anna, additional, Ruppert, Nadine, additional, Briefer, Elodie F., additional, Fuentes, Agustin, additional, and Hansen, Malene F., additional

Published
2024
Full Text
View/download PDF

25. Staleness-Reduction Mini-Batch K-Means

Author

Zhu, Xueying, Sun, Jie, He, Zhenhao, Jiang, Jiantong, and Wang, Zeke

Abstract

$K$ -means (km) is a clustering algorithm that has been widely adopted due to its simple implementation and high clustering quality. However, the standard km suffers from high computational complexity and is therefore time-consuming. Accordingly, the mini-batch (mbatch) km is proposed to significantly reduce computational costs in a manner that updates centroids after performing distance computations on just a mbatch, rather than a full batch, of samples. Even though the mbatch km converges faster, it leads to a decrease in convergence quality because it introduces staleness during iterations. To this end, in this article, we propose the staleness-reduction mbatch (srmbatch) km, which achieves the best of two worlds: low computational costs like the mbatch km and high clustering quality like the standard km. Moreover, srmbatch still exposes massive parallelism to be efficiently implemented on multicore CPUs and many-core GPUs. The experimental results show that srmbatch can converge up to $40\times $ – $130\times $ faster than mbatch when reaching the same target loss, and srmbatch is able to reach 0.2%–1.7% lower final loss than that of mbatch.

Published
2024
Full Text
View/download PDF

36. Effect of uniaxial stress on magnetic property of laminated amorphous sheets up to kilohertz range.

Author

Zhu, Xueying, Li, Yongjian, Dou, Yu, and Gao, Jingming

Abstract

Amorphous alloys are widely used as the core material for motors and transformers, which are subjected to stresses during manufacturing and operation processes. For instance, the compressive stress can influence the magnetic properties of amorphous alloys which leads to local overheating in the device and reduce the reliability. A high-frequency magnetizer with a stress application unit is designed and constructed to investigate the magnetic properties of amorphous alloys up to kilohertz range under stress conditions. The amorphous alloy sheets will fracture under stress because of their poor ductility and brittleness, so this paper uses laminated amorphous alloy sheets as the sample. The hysteresis loops and loss characteristics of amorphous alloys under stress ranging from 20 MPa tensile stress to 20 MPa compressive stress at a frequency of 50 Hz to 4 kHz are investigated. This research can provide data support for the design of cores in electromagnetic devices. [ABSTRACT FROM AUTHOR]

Published
2024
Full Text
View/download PDF

45. Construction of spidroin coacervate microdroplets and regulation of their morphology.

Author

Nie, Lanheng, Tao, Zhengyu, Zhu, Xueying, Huang, Xin, and Liu, Xiaoman

Abstract

In recent years, developing artificial cells of higher complexity has emerged as being key to simulating advanced life behaviors, among which coacervate microdroplets are a promising kind of model artificial cell. Constructing simple coacervate systems in vitro which can subsequently achieve specific responses to environmental stimuli to form coacervate microdroplet communities are fundamental for studying the interactions between liquid–liquid phase separated molecules and the way such interactions determine material properties, composition and phase behavior. Herein, we propose a membrane-free artificial cell based on recombinant spidroin, NT2RepCT, which utilizes the complex structure of spidroin to provide coacervate microdroplets with a unique population morphology in response to environmental stimuli. By changing the environmental conditions such as protein concentration, pH and temperature, the coacervate microdroplets of single-type, regular adhesion-type and irregular adhesion-type were statistically generalized, and it is highlighted that the adhesion-type of coacervate microdroplets depended on the α-helical percentage, complex folding degree of spidroin and internally hydrophobic environment of the coacervate, while it was inversely proportional to the surface hydrophobic environment. Much more interesting, regulation of the non-enzymatic polymerization reaction of oligonucleotides was successfully achieved by adjusting the population morphology of coacervate microdroplets. [ABSTRACT FROM AUTHOR]

Published
2023
Full Text
View/download PDF

Catalog

Books, media, physical & digital resources
See catalog results