Cheng Chen

Cheng Chen

Auto-generated profile, based on publicly available information

Cheng Chen

University of Electronic Science and Technology of China, Chengdu, China

Research Focus & Metrics

Academic Career Length

16Years (2011-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
11 October 2022 in IEEE Transactions on Image Processing
Verified
Journal Article
Closed access
HSHeng Tao Shen
Heng Tao Shen
CCCheng Chen
Cheng Chen
PWPeng Wang
Peng Wang
LGLianli Gao
Lianli Gao
MWMeng Wang
Meng Wang
JSJingkuan Song
Jingkuan Song
Referring Expression Comprehension (REC) aims to localize an image region of a given object described by a natural-language expression. While promising performance has been demonstrated, existing REC algorithms make a strong assumption that training data feeding into a model are given upfront, which degrades its practicality for real-world scenarios. In this paper, we propose Continual Referring Expression Comprehension (CREC), a new setting for REC, where a model is learning on a stream of incoming tasks. In order to continuously improve the model on sequential tasks without forgetting prior learned knowledge and without repeatedly re-training from a scratch, we propose an effective baseline method named Dual Modular Memorization (DMM), which alleviates the problem of catastrophic forgetting by two memorization modules: Implicit-Memory and Explicit-Memory. Specifically, the former module aims to constrain drastic changes to important parameters learned on old tasks when learning a new task; while the latter module maintains a buffer pool to dynamically select and store representative samples of each seen task for future rehearsal. We create three benchmarks for the new CREC setting, by respectively re-splitting three widely-used REC datasets RefCOCO, RefCOCO+ and RefCOCOg into sequential tasks. Extensive experiments on the constructed benchmarks demonstrate that our DMM method significantly outperforms other alternatives, based on two popular REC backbones. We make the source code and benchmarks publicly available to foster future progress in this field: https://github.com/zackschen/DMM.

26 October 2023
Verified
Conference Paper
Open access
CCChen Cheng
Chen Cheng
JSJingkuan Song
Jingkuan Song
XZXiaosu Zhu
Xiaosu Zhu
JZJunchen Zhu
Junchen Zhu
LGLianli Gao
Lianli Gao
HSHengtao Shen
Hengtao Shen
The focus of this study is on Unsupervised Continual Learning (UCL), as it presents an alternative to Supervised Continual Learning which needs high-quality manual labeled data. The experiments under UCL paradigm indicate a phenomenon where the results on the first few tasks are suboptimal. This phenomenon can render the model inappropriate for practical applications. To address this issue, after analyzing the phenomenon and identifying the lack of diversity as a vital factor, we propose a method named Codebook for Unsupervised Continual Learning (CUCL) which promotes the model to learn discriminative features to complete the class boundary. Specifically, we first introduce a Product Quantization to inject diversity into the representation and apply a cross quantized contrastive loss between the original representation and the quantized one to capture discriminative information. Then, based on the quantizer, we propose a effective Codebook Rehearsal to address catastrophic forgetting. This study involves conducting extensive experiments on CIFAR100, TinyImageNet, and MiniImageNet benchmark datasets. Our method significantly boosts the performances of supervised and unsupervised methods. For instance, on TinyImageNet, our method led to a relative improvement of 12.76% and 7% when compared with Simsiam and BYOL, respectively. Codes are publicly available at https://github.com/zackschen/CUCL

6 December 2011 in Procedia Engineering
Verified
Journal Article
Open access
PDF file
FZFengyuan Zhang
Fengyuan Zhang
ZLZhao Li
Zhao Li
MWMengli ChenLili Wen
Mengli ChenLili Wen
XJXiaolu Jia
Xiaolu Jia
CCCheng Chen
Cheng Chen
LZW algorithm is adopted in this paper to compress the bridge vibration data. Based on the data structure of chained lists, an improved version of the algorithm, in which a new method called forward-moving on frequently-used entries is applied, is implemented and optimized. It can be drawn that using the above-mentioned method can effectively improve the searching speed and save compression time. In addition, optimizing the location of forward-moved frequently-used entries and selecting the appropriate statistical frequency have better optimization effects. Finally, implemented with C language, the algorithm is proved to have high compression efficiency with less time cost in the practical compression of bridge vibration data.

8 January 2022 in Frontiers of Computer Science
Verified
Journal Article
Closed access
PDF file
CCCheng Chen
Cheng Chen
WZWeinan Zhang
Weinan Zhang
YYYong Yu
Yong Yu

15 July 2025 in Pattern Recognition
Verified
Journal Article
Closed access
CCCheng Chen
Cheng Chen
LGLianli Gao
Lianli Gao
PZPengpeng Zeng
Pengpeng Zeng
MCMingsheng Cao
Mingsheng Cao
HSHeng Tao Shen
Heng Tao Shen

1 October 2021 in arXiv
Verified
Preprint
Open access
PDF file
XFXue-Yong Fu
Xue-Yong Fu
CCCheng Chen
Cheng Chen
TLTahmid Rahman Laskar
Tahmid Rahman Laskar
STShashi Bhushan Tn
Shashi Bhushan Tn
SCSimon Corston-Oliver
Simon Corston-Oliver
Automatic Speech Recognition (ASR) systems generally do not produce punctuated transcripts. To make transcripts more readable and follow the expected input format for downstream language models, it is necessary to add punctuation marks. In this paper, we tackle the punctuation restoration problem specifically for the noisy text (e.g., phone conversation scenarios). To leverage the available written text datasets, we introduce a data sampling technique based on an n-gram language model to sample more training data that are similar to our in-domain data. Moreover, we propose a two-stage fine-tuning approach that utilizes the sampled external data as well as our in-domain dataset for models based on BERT. Extensive experiments show that the proposed approach outperforms the baseline with an improvement of 1:12% F1 score.