default author photo

Pangfeng Liu

affiliation not provided to SSRN

  • affiliation not provided to SSRN

SCHOLARLY PAPERS

2

DOWNLOADS

126

TOTAL CITATIONS

0

Scholarly Papers (2)

Actions:
Email selected abstracts
1.

Gpu Memory Usage Optimization for Backward Propagation in Deep Network Training

Number of pages: 29 Posted: 01 Feb 2024
Ning Wang, Pangfeng Liu, Ding-Yong Hong, Jan-Jan Wu and Tzu-Hsien Tsai
affiliation not provided to SSRN, affiliation not provided to SSRN, affiliation not provided to SSRN, Academia Sinica and affiliation not provided to SSRN
Downloads 93 (725,554)

Abstract:

Loading...

Deep Learning, Dynamic Programming, Memory usage optimization, Memory pressure, CheckPointing

2.

Hybrid Dual-Batch and Cyclic Progressive Learning for Efficient Distributed Training

Downloads 33 (1,277,121)

Hybrid Dual-Batch and Cyclic Progressive Learning for Efficient Distributed Training

Number of pages: 14 Posted: 27 Nov 2025
Kuan-Wei Lu, Ding-Yong Hong, Pangfeng Liu and Jan-Jan Wu
affiliation not provided to SSRN, affiliation not provided to SSRN, affiliation not provided to SSRN and Academia Sinica
Downloads 19 (1,513,226)

Abstract:

Loading...

Deep Learning, distributed training, Convolutional Neural Networks, parameter server, batch processing, progressive
learning

Hybrid Dual-Batch and Cyclic Progressive Learning for Efficient Distributed Training

Number of pages: 14 Posted: 25 Nov 2025
Kuan-Wei Lu, Ding-Yong Hong, Pangfeng Liu and Jan-Jan Wu
affiliation not provided to SSRN, affiliation not provided to SSRN, affiliation not provided to SSRN and Academia Sinica
Downloads 14 (1,570,077)

Abstract:

Loading...

Deep learning, distributed training, Convolutional Neural Networks, parameter server, batch processing, progressive
learning