• Annual Conference on Human and Language Technology
• /
• 2018.10a
• /
• pp.147-150
• /
• 2018

목적 지향 대화 시스템은 자연어 이해, 대화 관리자, 자연어 생성과 같은 세분화 모델들의 결합으로 이루어져있어 하위 모델에 대한 오류 전파에 취약하다. 이러한 문제점을 해결하기 위해 자연어 이해 모델과 대화 관리자를 하나의 네트워크로 구성하고 오류에 강건한 심층 Q 네트워크를 제안한다. 본 논문에서는 대화의 전체 흐름을 파악 할 수 있는 순환 신경망인 LSTM에 심층 Q 네트워크 적용한 심층 순환 Q 네트워크 기반 목적 지향 대화 시스템을 제안한다. 실험 결과, 제안한 심층 순환 Q 네트워크는 LSTM, 심층 Q 네트워크보다 각각 정밀도 1.0%p, 6.7%p 높은 성능을 보였다.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.32 no.9C
• /
• pp.810-819
• /
• 2007

The I/Q unbalance which is generated by non-ideal components such as a $90^{\circ}$ phase shifter and I/Q filters is an inevitable physical phenomenon and leads to performance degradation when we implement a coherent two-dimensional (2-D) modulation/demodulation system. This paper provides an exact and general expression for the SEP(symbol error probability) of DVB-S2 system with I/Q phase and amplitude unbalance over AWGN channel. Coordinate rotation and shift techniques used to redefine a received signal are key mathematical tools. In conclusion, the derived result is expressed as a linear combination of the 2-D Gaussian Q-functions.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 2002.12a
• /
• pp.162-165
• /
• 2002

Q-learning은 강화학습의 한 방법으로서, 여러 분야에 널리 응용되고 있는 기법이다. 최근에는 Linear Quadratic Regulation(이하 LQR) 문제에 성공적으로 적용된 바 있는데, 특히, 시스템모델의 파라미터에 대한 구체적인 정보가 없는 상태에서 적절한 입력과 출력만을 가지고 학습을 통해 문제를 해결할 수 있어서 상황에 따라서 매우 실용적인 대안이 될 수 있다. Neural Q-learning은 이러한 Q-learning의 Q-value를 MLP(multilayer perceptron) 신경망의 출력으로 대치시킴으로써, 비선형 시스템의 최적제어 문제를 다룰 수 있게 한 방법이다. 그러나, Neural Q방식은 신경망의 구조를 먼저 결정한 후 역전파 알고리즘을 이용하여 학습하는 절차를 취하기 때문에, 시행착오를 통하여 신경망 구조를 결정해야 한다는 점, 역전파 알고리즘의 적용으로 인해 신경망의 연결강도 값들이 지역적 최적해로 수렴한다는 점등의 문제점을 상속받는 한계가 있다. 따라서, 본 논문에서는 Neural-0 학습의 도구로, 역전파 알고리즘으로 학습되는 MLP 신경망을 사용하는 대신 최근 들어 여러 분야에서 그 성능을 인정받고 있는 서포트 벡터 학습법을 사용하는 방법을 택하여, $\varepsilon$-SVR(Epsilon Support Vector Regression)을 이용한 Q-value 근사 기법을 제안하고 관련 수식을 유도하였다. 그리고, 모의 실험을 통하여, 제안된 서포트 벡터학습 기반 Neural-Q 방법의 적용 가능성을 알아보았다.

• Journal of Internet Computing and Services
• /
• v.2 no.4
• /
• pp.83-90
• /
• 2001

Q+-reader, an E-Book reader system for XML based contents is introduced, The system is developed for the embedded platform, called Q+, which targets for the digital home electronics systems, The system supports XML contents and CSS based styling as described in Open e-book standard. The system renders the e-book contents for users, sc the contents are secured for a proper use. This will found an essential part of business infra for electronic books with copyright protection, The system was developed in Java language to be used in various platform;, and has an open architecture to be reusable for other standards as well.

• Smart Media Journal
• /
• v.4 no.3
• /
• pp.9-15
• /
• 2015

A pub(lish)/sub(scribe) system is a data forwarding system which forwards only interesting data among the whole published data, which is related to the subscriptions registered by end users. Classical pub/sub systems are realized by constructing a network of brokers which are responsible for storing or forwarding data. Along with the substantial increase of the population mobile users, it is required that the pub/sub system handles the subscriptions of user locations which changes continuously and frequently. In this paper, a new broker network based pub/sub system which efficiently handles the frequent changes of subscriptions related to user locations is proposed. In consideration of moving patterns of users and geographical property, the proposed pub/sub system categorize the entire data space into Slow Moving Region and Normal Moving Region, and manages the brokers which are responsible for these regions by using Q+R tree in order to handle user requests more efficiently. Through the extensive simulation, it is presented that the proposed Q+R tree based pub/sub system can reduce unnecessary needs of brokers and network traffic and can support the dynamic subscription related to user location.

• The Journal of Information Systems
• /
• v.32 no.1
• /
• pp.201-221
• /
• 2023

Purpose Customers consider the overall experience with the company as important as the quality of the product, and companies are also paying attention to creating long-term relationships with customers through optimal customer experiences. In this study, we propose a customer experience management process called 'CX-Q', which combines customer journey map and Q-methodology to understand the importance of customer experience based on the overall customer experience. Design/methodology/approach. CX-Q is a process that combines Q-methodology and customer journey maps, allowing stakeholders to explore and improve customer experiences at each contact point while engaging with brands, products, and services. It also enables them to derive customer experience insights and important management points for each segment. To demonstrate the usefulness of the proposed CX-Q, this study analyzed the experience of customers who used the Airbnb travel platform service as an example, applying the CX-Q process. Findings A total of four customer segments were derived, and it was found that each segment valued different attributes during the customer journey stage. The customer experience analysis using the CX-Q process proposed in this study is expected to help understand customers in more detail and assist in managing and improving customer experience.

• Journal of the Korean Institute of Illuminating and Electrical Installation Engineers
• /
• v.19 no.1
• /
• pp.148-154
• /
• 2005

This paper deals with performance degradations caused by RF I/Q impairments such as amplitude mismatch and phase mismatch in W-CDMA user equipment which uses QPSK(Quadrature Phase Shift Keying) modulation. The impacts of I/Q impairments on the BER(Bit Error Rate) are analyzed by using the variations of adjacent symbol distance. The BER versus amplitude mismatch and phase mismatch with QPSK constellation is reviewed through Matlab simulation. Performance degradation produced by RF I/Q impairments is measured with the implemented RF transceiver and modulation/demodulation test equipments through EVM(Error Vector Magnitude). The minimum performance requirements of amplitude mismatch and phase mismatch in W-CDMA user equipment are presented from the point of hardware implementation and the test method of the impairments is also included.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 2003.09a
• /
• pp.579-582
• /
• 2003

In this paper, we propose a 4-step inference method needed for constructing a natural language communication system. The method is used to obtain fuzzy quantifier Q′when QA is Fisr τ⇔ Q′(m′A) is mF is m"is τ is inferred (Q, Q′: quantifiers, A: fuzzy subject, m′, m": modifiers, y: fuzzy predicate, τ: truth qualifier). We show that Q′is resolved step by step for two types of Q, including a non-increasing type (few,...) and a non-decreasing type(most,...).

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 2003.05a
• /
• pp.85-88
• /
• 2003

강화학습의 한가지 방법인 Q-learning은 최근에 Linear Quadratic Regulation(이하 LQR) 문제에 성공적으로 적용된 바 있다. 특히, 시스템 모델의 파라미터에 대한 구체적인 정보없이 적절한 입ㆍ출력만으로 학습을 통해 문제의 해결이 가능하므로 상황에 따라 매우 실용적인 방법이 될 수 있다. 뉴럴-큐 기법은 이러한 Q-learning의 Q-value를 MLP(multilayer perceptron) 신경망의 출력으로 대치시켜, 비선형 시스템의 최적제어 문제를 다룰 수 있게 한 방법이다. 그러나, 뉴럴-큐 기법은 신경망의 구조를 먼저 결정한 후 역전파 알고리즘을 이용해 학습하는 절차를 행하므로, 시행착오를 통해 신경망 구조를 결정해야 한다는 점, 역전파 알고리즘의 적용에 따라 신경망의 연결강도 값들이 지역적 최적해로 수렴한다는 점등의 문제점이 있다. 본 논문에서는 뉴럴-큐 학습의 도구로 KFD회귀를 이용하여 Q 함수의 근사 기법을 제안하고 관련 수식을 유도하였다. 그리고, 모의 실험을 통하여, 제안된 뉴럴-큐 방법의 적용 가능성을 알아보았다.

• Journal of Institute of Control, Robotics and Systems
• /
• v.3 no.4
• /
• pp.350-356
• /
• 1997

It is desirable for autonomous robot systems to possess the ability to behave in a smooth and continuous fashion when interacting with an unknown environment. Although Q-learning requires a lot of memory and time to optimize a series of actions in a continuous state space, it may not be easy to apply the method to such a real environment. In this paper, for continuous state space applications, to solve problem and a triangular type Q-value model\ulcorner This sounds very ackward. What is it you want to solve about the Q-value model. Our learning method can estimate a current Q-value by its relationship with the neighboring states and has the ability to learn its actions similar to that of Q-learning. Thus, our method can enable robots to move smoothly in a real environment. To show the validity of our method, navigation comparison with Q-learning are given and visual tracking simulation results involving an 2-DOF SCARA robot are also presented.