Browsing by Author "Huang, Yu-Min"

Now showing 1 - 2 of 2

Stochastic Fusion for Multi-stream Neural Network in Video Classification
(2019-01-01)
Huang, Yu-Min
;
Tseng, Huan-Hsin
;
Chien, Jen-Tzung
;
電機工程學系
;
Department of Electrical and Computer Engineering
Spatial image and optical how provide complementary information for video representation and classification. Traditional methods separately encode two stream signals and then fuse them at the end of streams. This paper presents a new multi-stream recurrent neural network where streams are tightly coupled at each time step. Importantly, we propose a stochastic fusion mechanism for multiple streams of video data based on the Gumbel samples to increase the prediction power. A stochastic backpropagation algorithm is implemented to carry out a multi-stream neural network with stochastic fusion based on a joint optimization of convolutional encoder and recurrent decoder. Experiments on UCF101 dalaset illustrate the merits of the proposed stochastic fusion in recurrent neural network in terms of interpretation and classification performance.
Three-Dimensional Printed Porous Titanium Screw with Bioactive Surface Modification for Bone-Tendon Healing: A Rabbit Animal Model
(2020-05-01)
Huang, Yu-Min
;
Huang, Chih-Chieh
;
Tsai, Pei-I
;
Yang, Kuo-Yi
;
Huang, Shin-I
The interference screw fixation method is used to secure a graft in the tibial tunnel during anterior cruciate ligament reconstruction surgery. However, several complications have been reported, such as biodegradable screw breakage, inflammatory or foreign body reaction, tunnel enlargement, and delayed graft healing. Using additive manufacturing (AM) technology, we developed a titanium alloy (Ti6Al4V) interference screw with chemically calcium phosphate surface modification technology to improve bone integration in the tibial tunnel. After chemical and heat treatment, the titanium screw formed a dense apatite layer on the metal surface in simulated body fluid. Twenty-seven New Zealand white rabbits were randomly divided into control and additive manufactured (AMD) screw groups. The long digital extensor tendon was detached and translated into a tibial plateau tunnel (diameter: 2.0 mm) and transfixed with an interference screw while the paw was in dorsiflexion. Biomechanical analyses, histological analyses, and an imaging study were performed at 1, 3, and 6 months. The biomechanical test showed that the ultimate pull-out load failure was significantly higher in the AMD screw group in all tested periods. Micro-computed tomography analyses revealed early woven bone formation in the AMD screw group at 1 and 3 months. In conclusion, AMD screws with bioactive surface modification improved bone ingrowth and enhanced biomechanical performance in a rabbit model.