http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
Training-Free Hardware-Aware Neural Architecture Search with Reinforcement Learning
Linh Tam Tran,Sung-Ho Bae 한국방송·미디어공학회 2021 방송공학회논문지 Vol.26 No.7
Neural Architecture Search (NAS) is cutting-edge technology in the machine learning community. NAS Without Training (NASWOT) recently has been proposed to tackle the high demand of computational resources in NAS by leveraging some indicators to predict the performance of architectures before training. The advantage of these indicators is that they do not require any training. Thus, NASWOT reduces the searching time and computational cost significantly. However, NASWOT only considers high-performing networks which does not guarantee a fast inference speed on hardware devices. In this paper, we propose a multi objectives reward function, which considers the network’s latency and the predicted performance, and incorporate it into the Reinforcement Learning approach to search for the best networks with low latency. Unlike other methods, which use FLOPs to measure the latency that does not reflect the actual latency, we obtain the network’s latency from the hardware NAS bench. We conduct extensive experiments on NAS-Bench-201 using CIFAR-10, CIFAR-100, and ImageNet-16-120 datasets, and show that the proposed method is capable of generating the best network under latency constrained without training subnetworks.
Hoan Minh Tran,Tam Huu Nguyen,Viet Quoc Nguyen,Phuc Huynh Tran,Linh Duy Thai,Thuy Thu Truong,Le-Thu T. Nguyen,Ha Tran Nguyen 한국고분자학회 2019 Macromolecular Research Vol.27 No.1
The photoswitching poly(pyrene-1-ylmethyl-methacrylate-random-methyl methacrylate-random-methacrylate spirooxazine) was synthesized via atom transfer radical polymerization and characterized by proton nuclear magnetic resonance (1H NMR), gel permeation chromatography (GPC), Fourier transform infrared (FTIR) spectroscopy, UV-visible spectroscopy, and differential scanning calorimetry (DSC). The obtained copolymer exhibited the capability of erasable and rewritable photoimaging, making it a potential candidate for optical data storage materials. Moreover, the copolymer also showed the sensing ability for cyanide anions effect in aqueous solutions.
Determinants of Operational Self-Sustainability of Microfinance Institutions in Vietnam
LE, Thanh Tam,DAO, Lan Phuong,DO, Ngoc Mai,TRUONG, Thi Hoai Linh,NGUYEN, Thi Thuy Duong,TRAN, Chung Thuy Korea Distribution Science Association 2020 The Journal of Asian Finance, Economics and Busine Vol.7 No.10
The purpose of this paper is to investigate the determinants of the Operational Self-Sustainability (OSS) of Vietnamese microfinance institutions (MFIs). This research uses both qualitative and quantitative research methods: (i) qualitative research was via in-depth interviews with ten microfinance practitioners, policymakers and researchers; (ii) quantitative research was conducted by using panel data of 34 MFIs in the period 2011-2015 with binary logistics and OLS regressions. Results are as follows: (i) MFIs' OSS in Vietnam are mainly determined by five key factors: portfolio at risk (PAR>30), capital structure, gross loan portfolio, scope of activities and legal form; (ii) OSS are most affected by legal status (social organizations have better OSS than formal MFIs or programs/projects), location (MFIs focus in one province have higher OSS than working nationwide or just in one district), capital structure (MFIs with more equity proportion have higher OSS); (iii) surprisingly, average loan size per borrower and age of MFIs do not have statistically significant correlation with OSS. The key recommendations are: (i) MFIs should focus on its professionality and increase its equity; (ii) related stakeholders such as State Bank of Vietnam should promote the enabling ecosystem for microfinance development to enhance poverty reduction and economic development.
간략화된 MFCC를 이용한 PIR 센서 기반 침입감지 시스템
박원경(Won Gyeong Park),Tran Linh Tam,이호경(Ho Kyung Lee),조성원(Seongwon Cho) 한국지능시스템학회 2018 한국지능시스템학회논문지 Vol.28 No.4
PIR 센서를 사용하는 침입감지 시스템이 사람이 아닌 동물들로 인해 발생하는 오작동을 막기 위해 사람과 다른 동물을 구별하는 시스템의 필요성이 대두되고 있다. 본 논문에서는 사람과 동물을 구별하기 위해 PIR(Pyroelectric Infrared Sensor)센서로 부터 얻은 신호를 간략화된 MFCC(Mel-Frequency Cepstral Coefficients)를 이용하여 신호처리하여 성능을 개선한 방법을 제안한다. 음성신호처리에 주로 사용되는 MFCC를 사람과 동물신호를 구별하기에 적합하도록 수정하여 신호처리에 사용하였다. 처리된 신호들은 인공 신경회로망의 특징벡터로 사용되었다. 본 논문에서는 기존의 MFCC를 활용한 분류결과와 제안된 간략화 MFCC를 활용한 분류결과에 대해 비교하여 간략화된 MFCC가 정확도 개선에 우수함을 실험 결과로서 제시하였다. The intruder detection system using the PIR sensor has the problem that it can`t recognize human correctly. In this paper, we propose a new intruder detection system based on the PIR sensor to get around the drawbacks of the previous method. The signal captured using the PIR sensor is sampled, and its frequency feature is extracted using the simplified MFCC. The extracted features are used for the input of neural networks. After neural network is trained using various human and pet’s intrusion data, it is used for classifying human and pet in the intrusion situation. The experimental results show the superiority of the proposed method in comparison to the origianl MFCC.
Determinants Influencing Tax Compliance: The Case of Vietnam
NGUYEN, Thi Thuy Du'o'ng,PHAM, Thi My Linh,LE, Thanh Tam,TRUONG, Thi Hoai Linh,TRAN, Manh Dung Korea Distribution Science Association 2020 The Journal of Asian Finance, Economics and Busine Vol.7 No.2
The purpose of this paper is to ascertain the key factors affecting tax compliance among Vietnamese firms in Vietnam. We employ both qualitative and quantitative research methods. Qualitative research has been carried out through focus group discussions with ten chief accountants and tax officers. Quantitative research has been conducted through interviews with 200 firms (chief accountants or financial directors) in Vietnam. Analysis of the model includes the following stages: (i) Cronbach's test for reliability of the scale, (ii) exploratory factor analysis (EFA), (iii) confirmatory factor analysis (CFA), and (iv) structural equation model (SEM). The results of the research show that voluntary tax compliance is directly affected by the three factors of audit probability, corporate reputation and business ownership. The probability of audit and severity of sanctions have the strongest impact on tax compliance. Therefore, the tax authorities need to strengthen the inspection of tax declarations, tax payments and tax refunds of firms. The paper confirms that enforced tax compliance is directly affected by the three factors of audit probability, sanction severity and social norms. Voluntary compliance and compulsory compliance have an effect on tax compliance, though voluntary compliance has a more powerful impact.