• Asia pacific journal of information systems
• /
• 제26권1호
• /
• pp.124-142
• /
• 2016

Companies operating in social network platforms continue to improve and expand their marketing techniques. This study examines the practice of "gating", which involves virtual barriers between social network users and company content. Gates demand mandatory user endorsements, in the form of a Facebook "Likes", Twitter "retweets" etc., to gain access to company content, such as coupons and rewards,. Gating practices demand a mandatory endorsement before any content consumption takes place. Thus, while user endorsements are assumed to arise voluntarily from trusted known sources, gating practices would appear to violate this assumption. However, whether this violation lessens the effectiveness of gating practices still requires empirical validation. We investigate this question through the use of a unique panel data set that includes data on "like" endorsements obtained from a number of real-world Facebook business pages. Results of the study show that gating practices are effective for endorsement solicitation; however, gates may interfere with more traditional marketing activities.

• 한국지능시스템학회논문지
• /
• 제13권2호
• /
• pp.186-191
• /
• 2003

이동로봇(Mobile Robot)의 자율주행 기능에는 추종, 접근, 충돌회피, 경고 등의 여러 기능이 있다. 이 기능들을 하나의 Neural Network로 구성하고 학습하는 것은 쉬운 일이 아니다. 이동로봇의 자율주행 기능들을 각각의 Module로 구성하고 상황에 맞게 학습된 Module의 출력 값으로 이동로봇을 제어하면 단일 신경망의 단점을 보안할 수 있을 것이다. 이동로봇은 인간의 감각을 대신할 수 있는 다중 초음파 센서와 USB 카메라를 장착하고 있으며, 이곳에서 측정된 환경정보 데이터들은 Modular Neural Network(MNN)을 통해 학습을 한다. Expert Network(EN)의 활성화 함수를 최적결합으로 MNN을 구성하였고, 그 구조는 학습시간과 오차를 개선할 수 있을 것으로 본다. Gating Network(GN)는 MNN의 출력값인 이동로봇의 진행 방향과 속도를 스위칭 함으로써 제어하는 역할을 한다. 본 논문에서는 Modular Neural Network(MNN) 내의 Expert Network(EN)을 최적설계 하였고, 제안한 MNN의 검증을 위해 실시간으로 반복하여 이동로봇에 구현하였다. 그 실험의 결과값은 로봇을 상황에 맞게 운행, 제어하였고, 만족할 만한 성과를 얻을 수 있었다.

• Journal of Electrical Engineering and Technology
• /
• 제13권6호
• /
• pp.2479-2484
• /
• 2018

The filter bank multicarrier modulation (FBMC) technique is one of multicarrier modulation technique (MCM), which is mainly used to improve channel capacity of cognitive radio (CR) network and frequency spectrum access technique. The existing FBMC System contains serial to parallel converter, normal QAM modulation, Radix2 inverse FFT, parallel to serial converter and poly phase filter. It needs high area, delay and power consumption. To further reduce the area, delay and power of FBMC structure, a new clock gating technique is applied in the QAM modulation, radix2 multipath delay commutator (R2MDC) based inverse FFT and unified addition and subtraction (UAS) based FIR filter with parallel asynchronous self time adder (PASTA). The clock gating technique is mainly used to reduce the unwanted clock switching activity. The clock gating is nothing but clock signal of flip-flops is controlled by gate (i.e.) AND gate. Hence speed is high and power consumption is low. The comparison between existing QAM and proposed QAM with clock gating technique is carried out to analyze the results. Conversely, the proposed inverse R2MDC FFT with clock gating technique is compared with the existing radix2 inverse FFT. Also the comparison between existing poly phase filter and proposed UAS based FIR filter with PASTA adder is carried out to analyze the performance, area and power consumption individually. The proposed FBMC with clock gating technique offers low power and high speed than the existing FBMC structures.

• 한국지능시스템학회:학술대회논문집
• /
• 한국퍼지및지능시스템학회 2003년도 ISIS 2003
• /
• pp.492-495
• /
• 2003

Fuzzy Min-Max Neural Network(FMMNN) is a powerful classifier, It has, however, some problems. Learning result depends on the presentation order of input data and the training parameter that limits the size of hyperbox. The latter problem affects the result seriously. In this paper, the new approach to alleviate that without loss of on-line learning ability is proposed. The committee machine is used to achieve the multi-resolution FMMNN. Each expert is a FMMNN with fixed training parameter. The advantages of small and large training parameters are used at the same time. The parameters are selected by performance and independence measures. The Decision of each expert is guided by the gating network. Therefore the regional and parametric divide and conquer scheme are used. Simulation shows that the proposed method has better classification performance.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2003년도 하계종합학술대회 논문집 Ⅲ
• /
• pp.1565-1568
• /
• 2003

In this paper, we propose modular network to solve difficult and complex problems that are seldom solved with multi-layer neural network. The structure of modular neural network in researched by Jacobs and Jordan is selected in this paper. Modular network consists of several expert networks and a gating network which is composed of single-layer neural network or multi-layer neural network. We propose modular network structure using recurrent neural network, since the state of the whole network at a particular time depends on an aggregate of previous states as well as on the current input. Finally, we show excellence of the proposed network compared with modular network.

• 한국지능시스템학회:학술대회논문집
• /
• 한국퍼지및지능시스템학회 2003년도 ISIS 2003
• /
• pp.694-697
• /
• 2003

In this paper, we propose modular network to solve difficult and complex problems that are seldom solved with Multi-Layer Neural Network(MLNN). The structure of Modular Neural Network(MNN) in researched by Jacobs and jordan is selected in this paper. Modular network consists of several Expert Networks(EN) and a Gating Network(CN) which is composed of single-layer neural network(SLNN) or multi-layer neural network. We propose modular network structure using Recurrent Neural Network(RNN), since the state of the whole network at a particular time depends on aggregate of previous states as well as on the current input. Finally, we show excellence of the proposed network compared with modular network.

• 전기학회논문지
• /
• 제56권1호
• /
• pp.88-99
• /
• 2007

In this paper, a novel new pulse power generator based on IGBT stacks is proposed for pulse power application. Because it can generate up to 60kV pulse output voltage without any step- up transformer or pulse forming network, it has advantages of fast rising time, easiness of pulse width variation and rectangular pulse shape. Proposed scheme consists of series connected 9 power stages to generate maximum 60kV output pulse and one series resonant power inverter to charge DC capacitor voltage. Each power stages are configured as 8 series connected power cells and each power cell generates up to 850VDC pulse. Finally pulse output voltage is applied using total 72 series connected IGBTs. To reduce component for gate power supply, a simple and robust gate drive circuit is proposed. For gating signal synchronization, full bridge invertor and pulse transformer generates on-off signals of IGBT gating with gate power simultaneously and it has very good characteristics of short circuit protection.

• 한국지능시스템학회:학술대회논문집
• /
• 한국퍼지및지능시스템학회 2002년도 추계학술대회 및 정기총회
• /
• pp.415-418
• /
• 2002

MR의 자율주행 기능에는 추종, 접근, 충돌회피, 경고 등의 여러 기능이 있다. 이 기능들을하나의 Neural Network로 학습시키는 것은 어려운 일이다. 이것을 보안하고자 기능들을 각각의 Module로 구성하여 상황에 맞게 학습된 Module의 출력 값으로 MR을 제어하였다 로봇은 인간의 감각을 대신할 수 있는 다중 초음파 센서와 PC 카메라를 장착하고 있으며, 이곳에서 측정된 환경정보 데이터들은 Modular Neural Network을 통해 학습이 이루어진다 MNN에서의 출력값은 Gating Network(GN)에서 로봇의 진행 방향과 속도를 스위칭 출력함으로서 MR을 제어하는데 사용된다. MNN 내 EN의 활성화 함수 최적결합을 통해 효과적인 MNN을 구성하였다. 본 논문에서는 Modular Neural Network의 Expert Network(EN)을 최적설계 하였고, 제안한 MNN의 검증을 위해 실시간으로 MR에 구현하였다.

• 한국정보통신학회논문지
• /
• 제15권6호
• /
• pp.1213-1220
• /
• 2011

본 논문에서는 분류자들 속에 중요도 정보를 삽입하여 동적 중요도 결정이 가능한 앙상블 시스템을 제안하였다. 그동안 앙상블 시스템에서 중요도는 훈련이 끝나고 결정된 중요도를 사용하였다. 한 번 결정된 중요도는 테스트 데이터에 상관없이 정적으로 사용되었다. 이 문제를 푸는 방법으로 관문 네트워크에서 구조적으로 계층을 두는 프로세스를 추가하여 동적 중요도 결정이 가능하게 하는 방법이 있지만 프로세스가 추가된다는 단점이 있다. 본 논문에서는 이런 추가적인 프로세스 없이 간단하게 동적 중요도 결정이 가능한 방법을 보여주고 구조적 변경 없이 기존의 시스템에 쉽게 적용할 수 있으며 AdaBoost보다 나은 성능을 보여주는 알고리즘을 제안한다.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제13권6호
• /
• pp.3182-3198
• /
• 2019

Vision-based Human Pose Estimation has been considered as one of challenging research subjects due to problems including confounding background clutter, diversity of human appearances and illumination changes in scenes. To tackle these problems, we propose to use a new multi-stage convolution machine for estimating human pose. To provide better heatmap prediction of body joints, the proposed machine repeatedly produces multiple predictions according to stages with receptive field large enough for learning the long-range spatial relationship. And stages are composed of various modules according to their strategic purposes. Pyramid stacking module and dilation module are used to handle problem of human pose at multiple scales. Their multi-scale information from different receptive fields are fused with concatenation, which can catch more contextual information from different features. And spatial and channel information of a given input are converted to gating factors by squeezing the feature maps to a single numeric value based on its importance in order to give each of the network channels different weights. Compared with other ConvNet-based architectures, we demonstrated that our proposed architecture achieved higher accuracy on experiments using standard benchmarks of LSP and MPII pose datasets.