• Journal of Internet Computing and Services
• /
• v.24 no.4
• /
• pp.37-45
• /
• 2023

Adaptive has gained significant attention in Education Technology (EdTech), with personalized learning experiences becoming increasingly important. Next-generation chatbots, including models like ChatGPT, are emerging in the field of education. These advanced tools show great potential for delivering personalized and adaptive learning experiences. This paper reviews previous research on adaptive learning and the role of chatbots in education. Based on this, the paper explores current and future chatbot technologies to propose a framework for using ChatGPT or similar chatbots in adaptive learning. The framework includes personalized design, targeted resources and feedback, multi-turn dialogue models, reinforcement learning, and fine-tuning. The proposed framework also considers learning attributes such as age, gender, cognitive ability, prior knowledge, pacing, level of questions, interaction strategies, and learner control. However, the proposed framework has yet to be evaluated for its usability or effectiveness in practice, and the applicability of the framework may vary depending on the specific field of study. Through proposing this framework, we hope to encourage learners to more actively leverage current technologies, and likewise, inspire educators to integrate these technologies more proactively into their curricula. Future research should evaluate the proposed framework through actual implementation and explore how it can be adapted to different domains of study to provide a more comprehensive understanding of its potential applications in adaptive learning.

• The Journal of Asian Finance, Economics and Business
• /
• v.9 no.1
• /
• pp.43-51
• /
• 2022

Platform lending or online lending, sometimes called peer-to-peer (P2P) lending, arose due to the digital revolution to meet people's requirements for simple fund borrowing. It quickly became an alternative to other traditional lending techniques, for example, loans banks. Along with the growth of P2P lending, several academics have investigated how information technology is used in financial services, emphasizing extended application methods. This study proposes an enhanced technology acceptance model (TAM) that investigates how consumers embrace P2P lending platforms by using quality of service and perceived risk as drivers of trust, relative advantage and compatibility as drivers of perceived usefulness. For the purpose of this study, we created a questionnaire, distributed it to clients of P2P lending platforms and fintech services in general in cities in Java, Indonesia. We received 290 replies to our questionnaire. The data was analyzed to test the hypotheses using structural equation modeling (SEM). The findings show that consumers' trust, relative advantage, perceived usefulness, and perceived ease of use in P2P lending platforms substantially affect their views toward adoption. The research's findings are useful for fine-tuning platform marketing strategies and putting strategic goals into action.

• Annual Conference on Human and Language Technology
• /
• 2021.10a
• /
• pp.453-456
• /
• 2021

대화형 에이전트가 일관성 없는 답변, 재미 없는 답변을 하는 문제를 해결하기 위하여 최근 페르소나 기반의 대화 분야의 연구가 활발히 진행되고 있다. 그러나 한국어로 구축된 페르소나 대화 데이터는 아직 구축되지 않은 상황이다. 이에 본 연구에서는 영어 원본 데이터에서 한국어로 번역된 데이터를 활용하여 최초의 페르소나 기반 한국어 대화 모델을 제안한다. 전처리를 통하여 번역 품질을 향상시킨 데이터에 사전 학습 된 한국어 모델인 KoBERT와 KoELECTRA를 미세조정(fine-tuning) 시킴으로써 모델에게 주어진 페르소나와 대화 맥락을 고려하여 올바른 답변을 선택하는 모델을 학습한다. 실험 결과 KoELECTRA-base 모델이 가장 높은 성능을 보이는 것을 확인하였으며, 단순하게 사용자의 발화만을 주는 것 보다 이전 대화 이력이 추가적으로 주어졌을 때 더 좋은 성능을 보이는 것을 확인할 수 있었다.

• ETRI Journal
• /
• v.45 no.5
• /
• pp.874-886
• /
• 2023

A joint resource-optimization scheme is investigated for nonorthogonal multiple access (NOMA)-enhanced scalable video coding (SVC) multicast in unmanned aerial vehicle (UAV)-assisted radio-access networks (RANs). This scheme allows a ground base station and UAVs to simultaneously multicast successive video layers in SVC with successive interference cancellation in NOMA. A video quality-maximization problem is formulated as a mixed-integer nonlinear programming problem to determine the UAV deployment and association, RAN spectrum allocation for multicast groups, and UAV transmit power. The optimization problem is decoupled into the UAV deployment-association, spectrum-partition, and UAV transmit-power-control subproblems. A heuristic strategy is designed to determine the UAV deployment and association patterns. An upgraded knapsack algorithm is developed to solve spectrum partition, followed by fast UAV power fine-tuning to further boost the performance. The simulation results confirm that the proposed scheme improves the average peak signal-to-noise ratio, aggregate videoreception rate, and spectrum utilization over various baselines.

• Journal of Powder Materials
• /
• v.31 no.2
• /
• pp.137-145
• /
• 2024

In order to predict the process window of laser powder bed fusion (LPBF) for printing metallic components, the calculation of volumetric energy density (VED) has been widely calculated for controlling process parameters. However, because it is assumed that the process parameters contribute equally to heat input, the VED still has limitation for predicting the process window of LPBF-processed materials. In this study, an explainable machine learning (xML) approach was adopted to predict and understand the contribution of each process parameter to defect evolution in Ti alloys in the LPBF process. Various ML models were trained, and the Shapley additive explanation method was adopted to quantify the importance of each process parameter. This study can offer effective guidelines for fine-tuning process parameters to fabricate high-quality products using LPBF.

• IMMUNE NETWORK
• /
• v.23 no.2
• /
• pp.12.1-12.17
• /
• 2023

Ssu72, a dual-specificity protein phosphatase, not only participates in transcription biogenesis, but also affects pathophysiological functions in a tissue-specific manner. Recently, it has been shown that Ssu72 is required for T cell differentiation and function by controlling multiple immune receptor-mediated signals, including TCR and several cytokine receptor signaling pathways. Ssu72 deficiency in T cells is associated with impaired fine-tuning of receptor-mediated signaling and a defect in CD4⁺ T cell homeostasis, resulting in immune-mediated diseases. However, the mechanism by which Ssu72 in T cells integrates the pathophysiology of multiple immune-mediated diseases is still poorly elucidated. In this review, we will focus on the immunoregulatory mechanism of Ssu72 phosphatase in CD4⁺ T cell differentiation, activation, and phenotypic function. We will also discuss the current understanding of the correlation between Ssu72 in T cells and pathological functions which suggests that Ssu72 might be a therapeutic target in autoimmune disorders and other diseases.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2024.01a
• /
• pp.51-54
• /
• 2024

본 논문에서는 온 디바이스 국방 AI를 위한 효율적인 학습 방법을 제안한다. 제안하는 방법은 모델 전체를 재학습하는 대신 필요한 부분만 세밀하게 조정하여 계산 비용과 시간을 대폭 줄이는 PEFT 기법의 LoRa를 적용하였다. LoRa는 기존의 신경망 가중치를 직접 수정하지 않고 추가적인 낮은 랭크의 매트릭스를 학습하는 방식으로 기존 모델의 구조를 크게 변경하지 않으면서도, 효율적으로 새로운 작업에 적응할 수 있다. 또한 학습 파라미터 및 연산 입출력에 데이터에 대하여 32비트의 부동소수점(FP32) 대신 부동소수점(FP16, FP8) 또는 정수형(INT8)을 활용하는 경량화 기법인 양자화도 적용하였다. 적용 결과 학습시 요구되는 GPU의 사용량이 32GB에서 5.7GB로 82.19% 감소함을 확인하였다. 동일한 조건에서 동일한 데이터로 모델의 성능을 평가한 결과 동일 학습 횟수에선 LoRa와 양자화가 적용된 모델의 오류가 기본 모델보다 53.34% 증가함을 확인하였다. 모델 성능의 감소를 줄이기 위해서는 학습 횟수를 더 증가시킨 결과 오류 증가율이 29.29%로 동일 학습 횟수보다 더 줄어듬을 확인하였다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2024.05a
• /
• pp.774-776
• /
• 2024

The effective implementation of advanced speech recognition (ASR) systems necessitates the deployment of sophisticated keyword spotting models that are both responsive and resource-efficient. The initial local detection of user interactions is crucial as it allows for the selective transmission of audio data to cloud services, thereby reducing operational costs and mitigating privacy risks associated with continuous data streaming. In this paper, we address these needs and propose utilizing the Whisper-Tiny model with fine-tuning process to specifically recognize keywords from google speech dataset which includes 65000 audio clips of keyword commands. By adapting the model's encoder and appending a lightweight classification head, we ensure that it operates within the limited resource constraints of local devices. The proposed model achieves the notable test accuracy of 92.94%. This architecture demonstrates the efficiency as on-device model with stringent resources leading to enhanced accessibility in everyday speech recognition applications.

• Progress in Medical Physics
• /
• v.10 no.1
• /
• pp.17-22
• /
• 1999

INJ-I, INJ-E, PFN, BMI, and PRF were selected among the various factors which constitute a digital linear accelerator to find effects on the dose distribution by changing current and voltage within the permitted scale which Mevatron automatically maintained. We measured the absorbed dose using an ion chamber, analyzed the waveform of beam output using an oscilloscope, and measured symmetry and flatness using a dosimetry system. An RFA plus (Scanditronix, Sweden) device was used as a dosimetry system. Then an 0.6cc ion chamber (PR06C, USA), an electrometer (Capintec192, USA), and an oscilloscope (Tektronix, USA) were employed to measure the changes on the dose distribution characteristics by changing the beam-tuning parameters. When the currents and the voltages of INJ-I, INJ-E, PFN, BMI, and PRF were modified, we were able to see the notable change on the dose rate by examining the change of the output pulse using the oscilloscope and by measuring them using the ion chamber. However, the results of energy and flatness graph from RF A plus were almost identical. The factors had fine differences: INJ-I, INJ-E, PFN, BMI, and PRF had 0.01∼0.02% differences in D10/D20, 0.1∼0.2 % differences in symmetry, and 0.1∼0.4% differences in flatness. Since Mevatron controlled itself automatically to keep the reference value of the factor, it was not able to see large differences in the dose distribution. There were fine differences on the dose rate distribution when the voltage and the currents of the digitized factors were modified Nonetheless, a basic operational management information was achieved.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.3 no.2
• /
• pp.18-35
• /
• 1993

The water level control system of the steam generator in a pressurized water reactor and its control problems are analysed. In this work the stable control strategy during the low power operation and transient states is studied. To solve the problem, a fuzzy logic control method is applied as a basic algorithm of the controller. The control algorithm is based on the operator's knowledges and the experiences of manual operation for water level control at the compact nuclear simulator set up in Korea Atomic Energy Research Institute. From a viewpoint of the system realization, the control variables and rules are established considering simpler tuning and the input-output relation. The control strategy includes the dynamic tuning method and employs a substitutional information using the bypass valve opening instead of incorrectly measured signal at the low flow rate as the fuzzy variable of the flow rate during the pressure control mode of the steam generator. It also involves the switching algorithm between the control valves to suppress the perturbation of water level. The simulation results show that both of the fine control action at the small level error and the quick response at the large level error can be obtained and that the performance of the controller is improved.