• Journal of Broadcast Engineering
• /
• v.28 no.3
• /
• pp.302-313
• /
• 2023

This paper aims to develop a framework that can fully automate the quality management of training data used in large-scale Artificial Intelligence (AI) models built by the Ministry of Science and ICT (MSIT) in the 'AI Hub Data Dam' project, which has invested more than 1 trillion won since 2017. Autonomous driving technology using AI has achieved excellent performance through many studies, but it requires a large amount of high-quality data to train the model. Moreover, it is still difficult for humans to directly inspect the processed data and prove it is valid, and a model trained with erroneous data can cause fatal problems in real life. This paper presents a dataset reconstruction framework that removes abnormal data from the constructed dataset and introduces strategies to improve the performance of AI models by reconstructing them into a reliable dataset to increase the efficiency of model training. The framework's validity was verified through an experiment on the autonomous driving dataset published through the AI Hub of the National Information Society Agency (NIA). As a result, it was confirmed that it could be rebuilt as a reliable dataset from which abnormal data has been removed.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2021.05a
• /
• pp.240-243
• /
• 2021

Recently, there have been research results of applying Big data and AI technologies to the evaluation and individual learning for education. It is information technology innovations that collect dynamic and complex data, including student personal records, physiological data, learning logs and activities, learning outcomes and outcomes from social media, MOOCs, intelligent tutoring systems, LMSs, sensors, and mobile devices. In addition, e-learning was generated a large amount of learning data in the COVID-19 environment. It is expected that learning analysis and AI technology will be applied to extract meaningful patterns and discover knowledge from this data. On the learner's perspective, it is necessary to identify student learning and emotional behavior patterns and profiles, improve evaluation and evaluation methods, predict individual student learning outcomes or dropout, and research on adaptive systems for personalized support. This study aims to contribute to research in the field of education by researching and classifying machine learning technologies used in anomaly detection and recommendation systems for educational data.

• Information Systems Review
• /
• v.26 no.1
• /
• pp.19-56
• /
• 2024

This study is empirical research to enhance understanding of AI (artificial intelligence) training data project in South Korea. It primarily focuses on the various concerns regarding data quality from policy-executing institutions, data construction companies, and organizations utilizing AI training data to develop the most reliable algorithm for society. For academic contribution, this study suggests a theoretical foundation and research model for understanding AI training data quality and its antecedents, as well as the unique data and ethical aspects of AI. For this purpose, this study proposes a research model with important antecedents related to AI training data quality, such as data attribute factors, data building environmental factors, and data type-related factors. The study collects 393 sample data from actual practitioners and personnel from companies building artificial intelligence training data and companies developing artificial intelligence services. Data analysis was conducted through Fuzzy Set Qualitative Comparative Analysis (fsQCA) and Artificial Neural Network analysis (ANN), presenting academic and practical implications related to the quality of AI training data.

• Journal of the Korea Society of Computer and Information
• /
• v.28 no.1
• /
• pp.27-38
• /
• 2023

In this paper, we propose a conversational AI agent based on continual learning that can continuously learn and grow with new data over time. A continual learning-based conversational AI agent consists of three main components: Task manager, User attribute extraction, and Auto-growing knowledge graph. When a task manager finds new data during a conversation with a user, it creates a new task with previously learned knowledge. The user attribute extraction model extracts the user's characteristics from the new task, and the auto-growing knowledge graph continuously learns the new external knowledge. Unlike the existing conversational AI agents that learned based on a limited dataset, our proposed method enables conversations based on continuous user attribute learning and knowledge learning. A conversational AI agent with continual learning technology can respond personally as conversations with users accumulate. And it can respond to new knowledge continuously. This paper validate the possibility of our proposed method through experiments on performance changes in dialogue generation models over time.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• fall
• /
• pp.172-175
• /
• 2021

4차 산업혁명이 시작된 이래로 다양한 산업 분야에서 AI가 활용되고 있고, 그 중에서도 컴퓨터 비전 분야에서 딥러닝 기술이 각광받고 있다. 하지만 딥러닝 기술은 높은 전문 지식이 요구되어 관련 지식이 없는 일반인들은 활용하기 어렵다. 본 논문에서는 AI 관련 배경지식이 없는 사용자들도 UI를 통해 쉽게 이미지 분류 모델을 학습시킬 수 있는 노 코드 플랫폼에 관하여 기술하고, django 프레임워크를 이용해 웹 개발과 딥러닝 모델 학습을 통합 개발을 위한 아키텍처와 방향성을 제시하고자 한다. 사용자가 웹서버에 업로드한 이미지들을 웹 인터페이스를 통해 라벨링 하여 학습 데이터를 생성한 후, 이 데이터를 사용하여 모델을 학습시킨다. CNN 모델에 데이터를 학습시키는 과정과 생성된 모델 기반으로 이미지 예측하는 모듈을 통해 전문지식이 없는 사용자가 딥러닝 기술에 대해 쉽게 이해하고 이용하는 것을 기대할 수 있다.

• Annual Conference of KIPS
• /
• 2024.05a
• /
• pp.633-634
• /
• 2024

본 연구는 LSTM 모델이 수용 가능한 익명 보행자의 GPS 경로 범위와 훈련 데이터 셋의 크기에 대한 양상 분석을 목적으로 한다. 시계열 데이터인 GPS 경로 그리고 순환 신경망 LSTM 과 입력 구조를 이해하고, 두 가지 실험을 설계하여 LSTM 의 훈련 데이터 셋 수용을 파악한다. 실험에서는 장거리 데이터 셋을 학습한 모델과 그렇지 않은 모델을 비교하고, 훈련 데이터 셋 크기에 따른 학습 모델의 예측 값을 비교한다. 두 실험을 통해 GPS 경로 범위와 학습 가능한 경로의 가짓수에 대한 비교 분석 결과를 제시한다.

• Journal of The Korean Association of Information Education
• /
• v.24 no.1
• /
• pp.59-69
• /
• 2020

The purpose of this study is to design the framework of evaluation on learner's cognitive skill for artificial intelligence(AI) education through computational thinking. To design the rubric and framework for evaluating the change of leaner's intrinsic thinking, the evaluation process was consisted of a sequential stage with a) agency that cognitive learning assistance for data collection, b) abstraction that recognizes the pattern of data and performs the categorization process by decomposing the characteristics of collected data, and c) modeling that constructing algorithms based on refined data through abstraction. The evaluating framework was designed for not only the cognitive domain of learners' perceptions, learning, behaviors, and outcomes but also the areas of knowledge, competencies, and attitudes about the problem-solving process and results of learners to evaluate the changes of inherent cognitive learning about AI education. The results of the research are meaningful in that the evaluating framework for AI education was developed for the development of individualized evaluation tools according to the context of teaching and learning, and it could be used as a standard in various areas of AI education in the future.

• Annual Conference on Human and Language Technology
• /
• 2014.10a
• /
• pp.56-60
• /
• 2014

기계학습 방법에 기반한 자연어 분석은 학습 데이터가 필요하다. 학습 데이터가 구축된 소스 도메인이 아닌 다른 도메인에 적용할 경우 한국어 의미역 인식 기술은 15% 정도 성능 하락이 발생한다. 본 논문은 이러한 다른 도메인에 적용시 발생하는 성능 하락 현상을 극복하기 위해서 기존의 소스 도메인 학습 데이터를 활용하여, 소규모의 타겟 도메인 학습 데이터 구축만으로도 성능 하락을 최소화하기 위해 한국어 의미역 인식 기술에 prior 모델을 제안하며 기존의 도메인 적응 알고리즘과 비교 실험하였다. 추가적으로 학습 데이터에 사용되는 자질 중에서, 형태소 태그와 구문 태그의 자질 값을 기존보다 단순하게 적용하여 성능의 변화를 실험하였다.

• The Journal of the Convergence on Culture Technology
• /
• v.10 no.3
• /
• pp.909-914
• /
• 2024

This study suggests a quantitative evaluation of transfer learning, which is widely used in various AI fields, including image recognition for robot vision. Quantitative and qualitative analyses of results applying transfer learning are presented, but transfer learning itself is not discussed. Therefore, this study proposes a quantitative evaluation of transfer learning itself based on MNIST, a handwritten digit database. For the reference network, the change in recognition accuracy according to the depth of the transfer learning frozen layer and the ratio of transfer learning data and pre-training data is tracked. It is observed that when freezing up to the first layer and the ratio of transfer learning data is more than 3%, the recognition accuracy of more than 90% can be stably maintained. The transfer learning quantitative evaluation method of this study can be used to implement transfer learning optimized according to the network structure and type of data in the future, and will expand the scope of the use of robot vision and image analysis AI in various environments.

• Korean Journal of Remote Sensing
• /
• v.39 no.5_4
• /
• pp.1111-1123
• /
• 2023

In recent times, semantic image segmentation methods using deep learning models have been widely used for monitoring changes in surface attributes using remote sensing imagery. To enhance the performance of various UNet-based deep learning models, including the prominent UNet model, it is imperative to have a sufficiently large training dataset. However, enlarging the training dataset not only escalates the hardware requirements for processing but also significantly increases the time required for training. To address these issues, transfer learning is used as an effective approach, enabling performance improvement of models even in the absence of massive training datasets. In this paper we present three transfer learning models, UNet-ResNet50, UNet-VGG19, and CBAM-DRUNet-VGG19, which are combined with the representative pretrained models of VGG19 model and ResNet50 model. We applied these models to building extraction tasks and analyzed the accuracy improvements resulting from the application of transfer learning. Considering the substantial impact of learning rate on the performance of deep learning models, we also analyzed performance variations of each model based on different learning rate settings. We employed three datasets, namely Kompsat-3A dataset, WHU dataset, and INRIA dataset for evaluating the performance of building extraction results. The average accuracy improvements for the three dataset types, in comparison to the UNet model, were 5.1% for the UNet-ResNet50 model, while both UNet-VGG19 and CBAM-DRUNet-VGG19 models achieved a 7.2% improvement.