Article
Computer Science, Software Engineering
Budmonde Duinkharjav, Praneeth Chakravarthula, Rachel Brown, Anjul Patney, Qi Sun
Summary: This article explores the disconnect between human saccadic behaviors and spatial visual acuity through psychophysical studies. It develops a perceptual model that predicts temporal gaze behavior and validates the model using objective measurements and user studies. The article also demonstrates that sub-threshold image modifications commonly introduced in graphics pipelines can significantly alter human reaction timing.
ACM TRANSACTIONS ON GRAPHICS
(2022)
Article
Green & Sustainable Science & Technology
Guanjun Liu, Yun Wang, Hui Qin, Keyan Shen, Shuai Liu, Qin Shen, Yuhua Qu, Jianzhong Zhou
Summary: In this paper, a multi-network deep ensemble method is proposed to deal with the probabilistic prediction problems. The method can integrate different deep learning neural networks effectively and provide reliable uncertainty estimates for prediction. Furthermore, a spatiotemporal multi-network deep ensemble model is introduced, which employs advanced convolutional recurrent neural networks to capture spatiotemporal information and uses an intelligent optimization algorithm to assign weights to each network in the ensemble. An uncertainty quantification method is also introduced to provide reliable probability forecasts.
Article
Computer Science, Interdisciplinary Applications
Zhonghe Ren, Fengzhou Fang, Gaofeng Hou, Zihao Li, Rui Niu
Summary: In this study, a feature fusion method with multi-level information elements is proposed to improve the comprehensive performance of the appearance-based gaze estimation model. Experimental results show that optimizing the feature combination in the input control module and fine-tuning the computational architecture in the feature extraction module can improve the performance of the gaze estimation model, improving the performance and accessibility of the method.
JOURNAL OF COMPUTATIONAL DESIGN AND ENGINEERING
(2023)
Article
Energy & Fuels
Guanjun Liu, Hui Qin, Qin Shen, Hao Lyv, Yuhua Qu, Jialong Fu, Yongqi Liu, Jianzhong Zhou
Summary: A new spatiotemporal probabilistic prediction model combining a convolutional shared weight long short-term memory network and deep ensemble method is proposed in this paper to address solar radiation probabilistic prediction issues. The model optimizes uncertainty estimation and reliability of probabilistic prediction results by adjusting network structure and employing proper scoring rules. The model is shown to provide accurate point predictions, reasonable prediction intervals, and reliable probabilistic prediction results for a whole area in the evaluation against five state-of-the-art models and seven evaluation indicators.
Article
Engineering, Electrical & Electronic
Luming Zhang, Ming Chen, Guifeng Wang, Zhiming Wang
Summary: Accurately recognizing aerial photographs is a challenging task due to the asynchronous capture of low-resolution and high-resolution photos and the need for cross-resolution knowledge transfer. In this study, we propose a system that leverages low-resolution spatial composition and deep encoding to enhance the categorization of high-resolution aerial photographs.
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
(2022)
Article
Chemistry, Analytical
Hee Gyoon Kim, Ju Yong Chang
Summary: This study proposes a method for estimating gaze from arbitrary-sized low-resolution images by combining knowledge distillation and feature adaptation, which significantly improves gaze estimation performance.
Article
Social Issues
Quan-Hoang Vuong, Manh-Toan Ho, Minh-Hoang Nguyen, Thanh-Hang Pham, Thu-Trang Vuong, Quy Khuc, Hoang-Anh Ho, Viet-Phuong La
Summary: This study uses Animal Crossing: New Horizons as a case study to explore how video games can impact human environmental perceptions. Findings suggest that players tend to exploit the in-game environment despite their perceptions, indicating that simplified commercial game design may overlook opportunities to engage and educate players in pro-environmental activities.
TECHNOLOGY IN SOCIETY
(2021)
Article
Engineering, Chemical
Wei Fan, Shaojun Ren, Cong Yu, Haiquan Yu, Peng Wang, Fengqi Si
Summary: In modern industrial processes, there is an increasing focus on safety and reliability, leading to extensive research on process monitoring models. Considering multi-mode operating conditions simultaneously is crucial for process monitoring. This study proposes an efficient method, based on multi-mode probabilistic predictable feature analysis (MPPFA), for monitoring complex industrial processes. The method combines deep auto-encoder (DAE) and Gaussian mixture model (GMM) to extract low dimensional features and identify possible running modes. The effectiveness of the proposed method is demonstrated through experiments on a three-phase flow facility and a coal pulverising system.
JOURNAL OF THE TAIWAN INSTITUTE OF CHEMICAL ENGINEERS
(2023)
Article
Computer Science, Theory & Methods
Muhammad Ajmal Azad, Farhan Riaz, Anum Aftab, Syed Khurram Jah Rizvi, Junaid Arshad, Hany F. Atlam
Summary: This paper presents a novel approach called DEEPSEL for identifying malware and malicious codes in Android applications. DEEPSEL uses a set of features to characterize the behavior of applications and classify them as legitimate or malicious. Experimental results show that the proposed method achieves high accuracy and F-measure.
FUTURE GENERATION COMPUTER SYSTEMS-THE INTERNATIONAL JOURNAL OF ESCIENCE
(2022)
Article
Automation & Control Systems
Luming Zhang, Yongheng Shang, Ping Li, Hao Luo, Ling Shao
Summary: Computational photo quality evaluation is a useful technique in computer vision and graphics tasks. This article proposes a new community-aware photo quality evaluation framework that captures human perception using a latent topic model and multiple attributes. Experimental results show the competitiveness of our method and its high consistency with human gaze movements.
IEEE TRANSACTIONS ON CYBERNETICS
(2022)
Article
Green & Sustainable Science & Technology
Qiang Liu, Jianguang Xie, Fan Ding
Summary: This article introduces a data-driven feature-based learning application utilizing mobile phone data to detect segment traffic status. The application, evaluated through a large-scale field experiment using actual data in Jiangsu, China, performed well and can be considered as an emerging solution for traffic state monitoring.
Article
Multidisciplinary Sciences
Johan Nakuci, Jason Samaha, Dobromir Rahnev
Summary: Brain activity during a task shows significant variability. By using a data-driven clustering method, consistent EEG activity patterns across individual trials can be identified, revealing the different patterns associated with task characteristics. These findings highlight the across-trial variability in decision processes that are usually overlooked by experimenters, and provide a method for identifying endogenous brain state variability relevant to cognition and behavior.
Article
Medicine, General & Internal
Zibin Yang, Yuping Zhao, Jiarui Yu, Xiaobo Mao, Huaxing Xu, Luqi Huang
Summary: This study developed an intelligent tongue diagnosis system that utilizes deep learning on a mobile terminal for quick and accurate identification of the pathological features of the tongue, and proposed an efficient and accurate tongue image processing algorithm framework. The experimental results demonstrated satisfactory performance of the tongue diagnosis model with final classification accuracies of 93.33%, 89.60%, and 97.67% for tooth marks, spots, and fissures, respectively.
Article
Computer Science, Artificial Intelligence
Omar M. Elzeki, Mohamed Abd Elfattah, Hanaa Salem, Aboul Ella Hassanien, Mahmoud Shams
Summary: This study proposed a novel perceptual two-layer image fusion method using deep learning to obtain more informative CXR images from a COVID-19 dataset. Experimental results demonstrated the reliability of the algorithm in generating imbalanced COVID-19 datasets and the enriched features in the fused images. Through evaluation metrics, the proposed algorithm outperformed competitive image fusion algorithms in terms of image quality and characteristics.
PEERJ COMPUTER SCIENCE
(2021)
Article
Multidisciplinary Sciences
Ce Mo, Irene Cristofori, Guillaume Lio, Alice Gomez, Jean-Rene Duhamel, Chen Qu, Angela Sirigu
Summary: People tend to selectively trust others based on their appearance, and regardless of facial morphology, observers unconsciously increase the contrast of the eye area to make a face appear more trustworthy. Attraction judgements, however, depend on cultural processes.
Article
Computer Science, Information Systems
Luming Zhang, Jianwei Yin, Ping Li, Yongheng Shang, Roger Zimmermann, Ling Shao
IEEE TRANSACTIONS ON MULTIMEDIA
(2020)
Article
Computer Science, Information Systems
Li Yuan, Francis Eng Hock Tay, Ping Li, Jiashi Feng
IEEE TRANSACTIONS ON MULTIMEDIA
(2020)
Article
Computer Science, Artificial Intelligence
Ping Li, Qinghao Ye, Luming Zhang, Li Yuan, Xianghua Xu, Ling Shao
Summary: Video summarization using the proposed SUM-GDA method enhances diversity in summary frames by adapting global diverse attention mechanism, outperforming existing methods with remarkable improvements in computational efficiency.
PATTERN RECOGNITION
(2021)
Article
Computer Science, Information Systems
Ping Li, Chao Tang, Xianghua Xu
Summary: The proposed graph convolutional attention network (GCAN) for video summarization effectively integrates embedding learning and context fusion to generate compact and informative video summaries by considering both local and global relations among video frames.
FRONTIERS OF INFORMATION TECHNOLOGY & ELECTRONIC ENGINEERING
(2021)
Article
Computer Science, Artificial Intelligence
Ping Li, Pan Zhang, Xianghua Xu
Summary: This article introduces a GMMP (GCN Meta-learning with Multi-granularity POS) method based on multi-granularity POS for generating high-quality video captions. It models temporal dependency by treating video frames as nodes in a graph and captures POS information of words and phrases using a multi-granularity POS attention mechanism.
Article
Computer Science, Information Systems
Ping Li, Guopan Zhao, Xianghua Xu
Summary: This work presents a Coarse-to-Fine few-shot classification framework based on Metric-based Auxiliary learning to address the challenges of handling sample pairs with different similarity degrees and learning discriminant patterns from very few labeled samples per class.
INFORMATION SCIENCES
(2022)
Article
Computer Science, Artificial Intelligence
Ping Li, Guopan Zhao, Jiajun Chen, Xianghua Xu
Summary: Deep metric learning uses deep neural networks to learn the distance metric for data samples, aiming to encode the similarity between semantically related samples. However, learning a single metric using all samples fails to encode the similarity in different aspects. To address this issue, this paper proposes a Group Channel-wise Ensemble method that learns multiple distance metrics by partitioning the embedding space and using group channel-wise convolution blocks in convolution networks.
KNOWLEDGE-BASED SYSTEMS
(2023)
Article
Computer Science, Artificial Intelligence
Ping Li, Jiachen Cao, Xingchao Ye
Summary: This paper presents a point-level supervised temporal action detection framework based on prototype contrastive learning. It addresses the label sparsity and class imbalance problems by generating pseudo labels and utilizes prototype learning and contrastive representation learning to achieve discriminative prototype representations. Experimental results demonstrate the superior performance of the proposed method.
EXPERT SYSTEMS WITH APPLICATIONS
(2023)
Article
Computer Science, Information Systems
Ping Li, Pan Zhang, Tao Wang, Huaxin Xiao
Summary: A Time-Frequency recurrent Transformer with Diversity constraint (TFTD) is proposed for dense video captioning, which includes a time-frequency memory module to consider temporal relations and model motion dependency. The Determinantal Point Processes (DDP) is adopted to impose diversity loss and reduce redundancy in generated sentences. Experimental results demonstrate the superior performance of TFTD in terms of metrics and coherence compared to competitive alternatives.
INFORMATION PROCESSING & MANAGEMENT
(2023)
Article
Computer Science, Artificial Intelligence
Ping Li, Jiachen Cao, Li Yuan, Qinghao Ye, Xianghua Xu
Summary: In this paper, the authors propose a Multi-scale Dilation based Truncated Attention Proposal Network (MD-TAPN) model for temporal action detection, which achieves state-of-the-art performances on two benchmark databases. The model learns positive proposal relations by dynamically adjusting edge weights and suppresses disadvantageous relations by truncating negative attention scores. It also handles different action durations with a light multi-scale dilation module to increase proposal representation capacity.
PATTERN RECOGNITION
(2023)
Proceedings Paper
Computer Science, Artificial Intelligence
Qinghao Ye, Xiyue Shen, Yuan Gao, Zirui Wang, Qi Bi, Ping Li, Guang Yang
Summary: In this paper, a novel weakly supervised method for automated video highlight detection is proposed, achieving remarkable improvements over state-of-the-art methods. The method leverages audio-visual feature fusion, hierarchical temporal context encoding, and attention-gated instance aggregation to enhance detection performance and address existing issues. Extensive experiments on benchmark datasets demonstrate the effectiveness of the proposed approach.
2021 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV 2021)
(2021)
Proceedings Paper
Engineering, Electrical & Electronic
Peizheng Li, Han Cui, Aftab Khan, Usman Raza, Robert Piechocki, Angela Doufexi, Tim Farnham
Summary: This study introduces a WiFi indoor localisation technique based on deep learning, achieving high accuracy in different environments and exploring the effectiveness of model transfer to save training time and parameters.
2021 IEEE RADAR CONFERENCE (RADARCONF21): RADAR ON THE MOVE
(2021)
Article
Computer Science, Artificial Intelligence
Hongzhi Shi, Yong Li, Hancheng Cao, Xiangxin Zhou, Chao Zhang, Vassilis Kostakos
Summary: This paper introduces a novel semantics-aware mobility model that leverages large-scale semantic-rich spatial-temporal data from location-based social networks to capture human mobility motivation.
IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING
(2021)
Proceedings Paper
Computer Science, Information Systems
Huajie Shao, Dachun Sun, Jiahao Wu, Zecheng Zhang, Aston Zhang, Shuochao Yao, Shengzhong Liu, Tianshi Wang, Chao Zhang, Tarek Abdelzaher
WEB CONFERENCE 2020: PROCEEDINGS OF THE WORLD WIDE WEB CONFERENCE (WWW 2020)
(2020)
Proceedings Paper
Computer Science, Information Systems
Yu Meng, Jiaxin Huang, Guangyuan Wang, Zihan Wang, Chao Zhang, Yu Zhang, Jiawei Han
WEB CONFERENCE 2020: PROCEEDINGS OF THE WORLD WIDE WEB CONFERENCE (WWW 2020)
(2020)