Continuous Timescale Long-Short Term Memory Neural Network for Human Intent Understanding

Yu, Zhibin,Moirangthem, Dennis S.,Lee, Minho Frontiers Media S.A. 2017 Frontiers in neurorobotics Vol.11 No.-

<P>Understanding of human intention by observing a series of human actions has been a challenging task. In order to do so, we need to analyze longer sequences of human actions related with intentions and extract the context from the dynamic features. The multiple timescales recurrent neural network (MTRNN) model, which is believed to be a kind of solution, is a useful tool for recording and regenerating a continuous signal for dynamic tasks. However, the conventional MTRNN suffers from the vanishing gradient problem which renders it impossible to be used for longer sequence understanding. To address this problem, we propose a new model named Continuous Timescale Long-Short Term Memory (CTLSTM) in which we inherit the multiple timescales concept into the Long-Short Term Memory (LSTM) recurrent neural network (RNN) that addresses the vanishing gradient problem. We design an additional recurrent connection in the LSTM cell outputs to produce a time-delay in order to capture the slow context. Our experiments show that the proposed model exhibits better context modeling ability and captures the dynamic features on multiple large dataset classification tasks. The results illustrate that the multiple timescales concept enhances the ability of our model to handle longer sequences related with human intentions and hence proving to be more suitable for complex tasks, such as intention recognition.</P>

Sensitivity Analysis of Laser Quenching Parameters of ASTM 1045 of Disk Laser Based on Response Surface Method

Zhibin Yu,Chang Li,Zhengwei Chen,Yunfei Li,Xing Han 대한금속·재료학회 2021 METALS AND MATERIALS International Vol.27 No.5

Laser hardening is an important branch of laser surface hardening technology, which is widely used in metallurgy, transportation,machinery manufacturing, aerospace and other fields. At present, relying entirely on experience or process trial-anderrormethod, it can not effectively reveal the transient mechanism of laser quenching of disk laser, which is not conduciveto shortening the research and development cycle and saving costs. The numerical simulation provides an effective way toobtain the dynamic evolution law of multi-field coupling in laser quenching process. In this paper, a thermo-mechanicalcoupling model of ASTM 1045 laser quenching process by disk laser is established. In the model, the temperature-dependentphysical parameters were calculated by CALPHAD method. The transient law of temperature and microstructural transformationduring quenching was obtained by solving the model. The formation and transformation degree of martensite werecharacterized by the dynamic changes of the depth and width of quenched transformation layer. The quenching structure andtransformation hardening law were observed by Axioskop 2 SEM and Zeiss-IGMA HD FE-SEM to verify the accuracy ofthe simulation results. On this basis, the process parameters of laser quenching were sampled by Monte-Carlo method basedon response surface methodology. The sensitivity effects of different process parameters on the temperature field and phasechange field of laser quenching were analyzed, which laid a theoretical foundation for the optimization of process parameters.

Traffic Classification based on Adjustable Convex-hull Support Vector Machines

위즈빈(Zhibin Yu),최용도(Yong-Do Choi),길기범(Gi-Beom Kil),김승호(Sung-Ho Kim) 한국컴퓨터정보학회 2012 韓國컴퓨터情報學會論文誌 Vol.17 No.3

트래픽 분류는 트래픽 관리하는데 중요한 역할을 차지하고 있다. 전통적인 방법은 P2P와 암호화 트래픽을 제대로 분류할 수 없는 문제가 있다. 서포트 벡터 머신은 기존의 문제를 해결할 수 있고 병목 현상을 극복할 수 있는 유용한 분류 도구이다. 하지만 서포트 벡터 머신의 주요 장점은 이차 프로그래밍(QP)문제 때문에 큰 데이터 집단을 훈련하는데 시간을 소모한다. 그러나 유용한 서포트 벡터는 전체 데이터에서 극히 일부분이다. 만약 우리가 훈련전에 쓸모없는 벡터들을 삭제할 수 있다면, 시간을 절약하고 정확도를 유지할 수 있다. 이 논문에서 우리는 대규모 데이터를 다룰 때 훈련 속도를 빠르게 하기위해 순차적인 방법을 통해 쓸모없는 벡터들을 제거하기 위한 가능성을 논의하였다. Traffic classification plays an important role in traffic management. To traditional methods, P2P and encryption traffic may become a problem. Support Vector Machine (SVM) is a useful classification tool which is able to overcome the traditional bottleneck. The main disadvantage of SVM algorithms is that it's time-consuming to train large data set because of the quadratic programming (QP) problem. However, the useful support vectors are only a small part of the whole data. If we can discard the useless vectors before training, we are able to save time and keep accuracy. In this article, we discussed the feasibility to remove the useless vectors through a sequential method to accelerate training speed when dealing with large scale data.

High optical quality polycrystalline indium phosphide grown on metal substrates by metalorganic chemical vapor deposition

Zheng, Maxwell,Yu, Zhibin,Joon Seok, Tae,Chen, Yu-Ze,Kapadia, Rehan,Takei, Kuniharu,Aloni, Shaul,Ager, Joel W.,Wu, Ming,Chueh, Yu-Lun,Javey, Ali American Institute of Physics 2012 JOURNAL OF APPLIED PHYSICS - Vol.111 No.12

Analysis of the extrusion pressure of a cylindrical extruder for extruding highly viscous fluids

Zhibin Sun,Baojun Shen,Yu He,Jiecai Long,Xiaobin Zhan,Yujin Li,Xiwen Li 한국화학공학회 2022 Korean Journal of Chemical Engineering Vol.39 No.10

Extrusion pressure is crucial for the security and performance of a cylindrical extruder during the extrusion process. In this study, a validated CFD model was adopted to evaluate the relationship between the extrusion velocity, fluid viscosity, and the extrusion pressure of a cylindrical extruder while extruding highly viscous fluids. The simulated and experimental results of the extrusion pressure and velocity profiles show good agreement. This study reveals that extrusion pressure evolution can be divided into two stages during the extrusion process. At stage I, the distance between the ram and the bottom of the vessel (liquid height) is greater than the critical height and the extrusion pressure remains almost constant. At stage II, the distance is less than the critical height and the extrusion pressure increases exponentially. The results indicate that an increase in extrusion velocity and fluid viscosity leads to a linear increase in the extrusion pressure at stage I. Furthermore, by introducing a pressure number, Np, and a pressurerelated Reynolds number, Rep, a novel correlation of the extrusion pressure with the extrusion velocity, viscosity of highly viscous fluids and liquid height has been developed.

Numerical investigation of the mixing process in a Twin Cam Mixer: Influence of triangular cam height-base ratio and eccentricity

Yu He,Xiwen Li,Jiecai Long,Baojun Shen,Zhibin Sun,Yili Yang,Xiaobin Zhan 한국화학공학회 2021 Korean Journal of Chemical Engineering Vol.38 No.3

The twin cam mixer (TCM), as a general-purpose mixer, shares many attributes in common with 3D industrial mixers, like the internal mixer. We investigated the mixing process in a 2D TCM with two identical isosceles triangular cams rotating at 0.5 rpm. A 2D numerical model coupled with the species transport model was employed to study the influence of cam height-base ratio and eccentricity qualitatively and quantitatively, and both were found to have a significant effect on the mixing behavior of the mixer. Furthermore, a dimensionless parameter, named the modified pressurization coefficient, is put forward to quantify the geometry of the mixer. The logarithmic relationship between the modified pressurization coefficient and the mixing quality was discovered and expected to provide new ideas for establishing the relationship between the geometric parameters of a mixer and its mixing performance.

Understanding human intention by connecting perception and action learning in artificial agents

Kim, Sangwook,Yu, Zhibin,Lee, Minho Elsevier 2017 Neural networks Vol.92 No.-

<P><B>Abstract</B></P> <P>To develop an advanced human–robot interaction system, it is important to first understand how human beings learn to perceive, think, and act in an ever-changing world. In this paper, we propose an intention understanding system that uses an Object Augmented-Supervised Multiple Timescale Recurrent Neural Network (OA-SMTRNN) and demonstrate the effects of perception–action connected learning in an artificial agent, which is inspired by psychological and neurological phenomena in humans. We believe that action and perception are not isolated processes in human mental development, and argue that these psychological and neurological interactions can be replicated in a human–machine scenario. The proposed OA-SMTRNN consists of perception and action modules and their connection, which are constructed of supervised multiple timescale recurrent neural networks and the deep auto-encoder, respectively, and connects their perception and action for understanding human intention. Our experimental results show the effects of perception–action connected learning, and demonstrate that robots can understand human intention with OA-SMTRNN through perception–action connected learning.</P>

LasR Might Act as an Intermediate in Overproduction of Phenazines in the Absence of RpoS in Pseudomonas aeruginosa

( Qiuning He ),( Zhibin Feng ),( Yanhua Wang ),( Kewen Wang ),( Kailu Zhang ),( Le Kai ),( Xiuying Hao ),( Zhifen Yu ),( Lijuan Chen ),( Yihe Ge ) 한국미생물생명공학회(구 한국산업미생물학회) 2019 Journal of microbiology and biotechnology Vol.29 No.8

As an opportunistic bacterial pathogen, Pseudomonas aeruginosa PAO1 contains two phenazineproducing gene operons, phzA1B1C1D1E1F1G1 (phz1) and phzA2B2C2D2E2F2G2 (phz2), each of which is independently capable of encoding all enzymes for biosynthesizing phenazines, including phenazine-1-carboxylic acid and its derivatives. Other previous study reported that the RpoS-deficient mutant SS24 overproduced pyocyanin, a derivative of phenazine-1- carboxylic acid. However, it is not known how RpoS mediates the expression of two phz operons and regulates pyocyanin biosynthesis in detail. In this study, with deletion of the rpoS gene in the PAΔphz1 mutant and the PAΔphz2 mutant respectively, we demonstrated that RpoS exerted opposite regulatory roles on the expression of the phz1and phz2 operons. We also confirmed that the phz1 operon played a critical role and especially biosynthesized much more phenazines than the phz2 operon when the rpoS gene was knocked out in P. aeruginosa. By constructing the translational reporter fusion vector lasR’-’lacZ and the chromosomal fusion mutant PAΔlasR::lacZ, we verified that RpoS deficiency caused increased expression of lasR, a transcription regulator gene in a first quorum sensing system (las) that activates overexpression of the phz1 operon, suggesting that in the absence of RpoS, LasR might act as an intermediate in overproduction of phenazine biosynthesis mediated by the phz1 operon in P. aeruginosa.

IEEE 802.1AS 에서 균형적인 평균 전달 지연 시간을 고려한 시간 동기화 방법

강성환 ( Sunghwan Kang ),위즈빈 ( Zhibin Yu ),안정균 ( Jeong-gyun Ahn ),김승호 ( Sung-ho Kim ) 한국정보처리학회 2009 한국정보처리학회 학술대회논문집 Vol.16 No.2

네트워크를 이용하는 장치들간의 시간 동기는 최근 중요한 이슈로 떠오르고 있다. 이들 장치들간의 시간 동기를 위해 IEEE 802.1AS 에서 규격이 작성되고 있다. 네트워크 전체에 시간을 동기화하기 위해 가장 정확한 시간을 가진 장치를 선정하는 단계, 이웃하는 장치들간의 평균 전달 지연 시간을 측정하는 단계, 기준 시간에 모든 장치들이 시간을 동기화 하는 단계를 거치게 된다. 본 논문에서는 균형적인 평균 전달지연시간 측정하고 이를 시간 동기화에 적용하는 방법을 제안한다.

Protective effect and mechanism of ginsenoside Rg2 on atherosclerosis

Qianqian Xue,Tao Yu,Zhibin Wang,Xiuxiu Fu,Xiaoxin Li,Lu Zou,Min Li,Jae Youl Cho,Yanyan Yang The Korean Society of Ginseng 2023 Journal of Ginseng Research Vol.47 No.2

Background: Ginsenoside Rg2 (Rg2) has a variety of pharmacological activities and provides benefits during inflammation, cancer, and other diseases. However, there are no reports about the relationship between Rg2 and atherosclerosis. Methods: We used 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) to detect the cell viability of Rg2 in vascular smooth muscle cells (VSMCs) and human umbilical vein endothelial cells (HUVECs). The expression of inflammatory factors in HUVECs and the expression of phenotypic transformation-related marker in VSMCs were detected at mRNA levels. Western blot method was used to detect the expression of inflammation pathways and the expression of phenotypic transformation at the protein levels. The rat carotid balloon injury model was performed to explore the effect of Rg2 on inflammation and phenotypic transformation in vivo. Results: Rg2 decreased the expression of inflammatory factors induced by lipopolysaccharide in HUVECs-without affecting cell viability. These events depend on the blocking regulation of NF-κB and p-ERK signaling pathway. In VSMCs, Rg2 can inhibit the proliferation, migration, and phenotypic transformation of VSMCs induced by platelet derived growth factor-BB (PDGF-BB)-which may contribute to its anti-atherosclerotic role. In rats with carotid balloon injury, Rg2 can reduce intimal proliferation after injury, regulate the inflammatory pathway to reduce inflammatory response, and also suppress the phenotypic transformation of VSMCs. Conclusion: These results suggest that Rg2 can exert its anti-atherosclerotic effect at the cellular level and animal level, which provides a more sufficient basis for ginseng as a functional dietary regulator.