Search

Your search keyword '"Yu Hen Hu"' showing total 931 results
Start Over Author "Yu Hen Hu"

Refine your results

more »

more »

more »
931 results on '"Yu Hen Hu"'

101. Enhancing the Revised NIOSH Lifting Equation using ComputerVision

Author

Ming-Lun Lu, Runyu L. Greene, Yu Hen Hu, Guanhua Chen, and Robert G. Radwin

Subjects
Medical Terminology, Computer science, Medical Assisting and Transcription
Abstract

Trunk kinematics (i.e. speed and acceleration), a risk factor associated with low back pain (LBP), is difficult to measure in the field and is not incorporated in popular lifting analysis tools. In this study, computationally efficient computer vision-based methods were used to estimate trunk kinematics from video recordings collected in a previous prospective study. We explored the relationships between trunk kinematics and health outcomes, and if incorporating trunk kinematics into the revised NIOSH Lifting Equation (RNLE) may improve its predictability of LBP risk. Significant correlations between the percentage of LBP and the average trunk speed and acceleration were found. A multivariate logistic regression model was constructed using the kinematics variables and the RNLE lifting index (LI) as independent variables to predict the probability that a task was associated with high LBP risk. When trunk kinematics was incorporated to the LI model, the predictability of the model improved significantly ( p=0.003).

Published
2021

102. A Mixture of Experts Approach for Low-Cost DNN Customization

Author

Boyu Zhang, Yu Hen Hu, and Azadeh Davoodi

Subjects
ComputingMilieux_THECOMPUTINGPROFESSION, Edge device, biology, Artificial neural network, business.industry, Computer science, Feature extraction, Machine learning, computer.software_genre, biology.organism_classification, GeneralLiterature_MISCELLANEOUS, Data modeling, Personalization, Chen, Hardware and Architecture, Overhead (computing), Artificial intelligence, Electrical and Electronic Engineering, Performance improvement, business, computer, Software
Abstract

Editor’s notes: This article introduces a novel architecture for local learning, namely mixture of experts (MoEs), to customize a deep neural network (DNN) that is deployed on an edge device with performance improvement and low implementation overhead. — Deming Chen, University of Illinois at Urbana-Champaign

Published
2021

142. Multi-Image Encryption Based on Compressed Sensing and Deep Learning in Optical Gyrator Domain

Author

Jun Wang, Renjie Ni, Fan Wang, and Yu Hen Hu

Subjects
Computer science, ComputingMethodologies_IMAGEPROCESSINGANDCOMPUTERVISION, Chaotic, Data_CODINGANDINFORMATIONTHEORY, 02 engineering and technology, Iterative reconstruction, Encryption, Multi-image encryption, 01 natural sciences, 010309 optics, Gyrator, Matrix (mathematics), Computer Science::Multimedia, 0103 physical sciences, Ciphertext, 0202 electrical engineering, electronic engineering, information engineering, Applied optics. Photonics, Electrical and Electronic Engineering, compressed sensing, Computer Science::Cryptography and Security, business.industry, deep learning, Plaintext, QC350-467, Optics. Light, Atomic and Molecular Physics, and Optics, optical gyrator transform, TA1501-1820, Compressed sensing, 020201 artificial intelligence & image processing, business, Algorithm
Abstract

In this paper, a multi-image encryption scheme based on compressed sensing (CS) and deep learning in the optical gyrator domain is proposed. Firstly, multiple plaintext images are compressed by CS to obtain multiple measurements, and then the pixels of each measurement are scrambled by using a chaotic system. Secondly, the scrambled measurements are combined into a matrix and diffused by XOR operation with a chaotic matrix. Finally, the diffused matrix is encoded with a random phase and an optical gyrator transform to obtain a complex-valued matrix, and the amplitude of the complex-valued matrix is taken as the ciphertext. In decrypt, plaintext images are reconstructed from the CS measurements by a neural network, which achieves high reconstruction speed and quality compared with the traditional algorithm. Especially, the data amount of plaintext images can be compressed by up to 8 times while achieving high decryption quality. To our best knowledge, CS reconstruction algorithms based on deep learning is firstly used for image encryption. Moreover, the proposed scheme is highly robust against occlusion, noise, and chosen-plaintext attack.

Published
2021

Catalog

Books, media, physical & digital resources
See catalog results