• 한국정보처리학회:학술대회논문집
• /
• 한국정보처리학회 2023년도 추계학술발표대회
• /
• pp.449-452
• /
• 2023

The rapid development of the Internet has led to a significant increase in multimedia content in social networks. How to better analyze and improve video classification models has become an important task. Deep learning models have typical "black box" characteristics. The model requires explainable analysis. This article uses two classification models: ConvLSTM and VGG16+LSTM models. And combined with the explainable method of LRP, generate visualized explainable results. Finally, based on the experimental results, the accuracy of the classification model is: ConvLSTM: 75.94%, VGG16+LSTM: 92.50%. We conducted explainable analysis on the VGG16+LSTM model combined with the LRP method. We found VGG16+LSTM classification model tends to use the frames biased towards the latter half of the video and the last frame as the basis for classification.

• International journal of advanced smart convergence
• /
• 제9권2호
• /
• pp.90-104
• /
• 2020

It is difficult for non-experts to build machine learning (ML) models at the level that satisfies their needs. Deep learning models are even more challenging because it is unclear how to improve the model, and a trial-and-error approach is not feasible since training these models are time-consuming. To assist these novice users, we examined how interactive and explainable feedback while training a deep learning network can contribute to model performance and users' satisfaction, focusing on the data preparation process. We conducted a user study with 31 participants without expertise, where they were asked to improve the accuracy of a deep learning model, varying feedback conditions. While no significant performance gain was observed, we identified potential barriers during the process and found that interactive and explainable feedback provide complementary benefits for improving users' understanding of ML. We conclude with implications for designing an interface for building ML models for novice users.

• 한국컴퓨터정보학회:학술대회논문집
• /
• 한국컴퓨터정보학회 2022년도 제66차 하계학술대회논문집 30권2호
• /
• pp.99-100
• /
• 2022

본 논문에서는 유방암 초음파 이미지를 학습한 multi-modal 구조를 이용하여 유방암을 판별하는 인공지능을 제안한다. 학습된 인공지능은 유방암을 판별과 동시에, 설명 가능한 인공지능 기법과 ROI를 함께 사용하여 종양의 위치를 나타내준다. 시각적으로 판단 근거를 제시하기 때문에 인공지능의 판단 신뢰도는 더 높아진다.

• 인터넷정보학회논문지
• /
• 제23권2호
• /
• pp.97-105
• /
• 2022

Web services show a rapid evolution and integration to meet the increased users' requirements. Thus, web services undergo updates and may have performance degradation due to undetected faults in the updated versions. Due to these faults, many performances and regression anomalies in web services may occur in real-world scenarios. This paper proposed applying the deep learning model and innovative explainable framework to detect performance and regression anomalies in web services. This study indicated that upper bound and lower bound values in performance metrics provide us with the simple means to detect the performance and regression anomalies in updated versions of web services. The explainable deep learning method enabled us to decide the precise use of deep learning to detect performance and anomalies in web services. The evaluation results of the proposed approach showed us the detection of unusual behavior of web service. The proposed approach is efficient and straightforward in detecting regression anomalies in web services compared with the existing approaches.

• 대한의용생체공학회:의공학회지
• /
• 제43권4호
• /
• pp.259-270
• /
• 2022

Artificial intelligence (AI) has been studied in various fields of medical imaging. Currently, top-notch deep learning (DL) techniques have led to high diagnostic accuracy and fast computation. However, they are rarely used in real clinical practices because of a lack of reliability concerning their results. Most DL models can achieve high performance by extracting features from large volumes of data. However, increasing model complexity and nonlinearity turn such models into black boxes that are seldom accessible, interpretable, and transparent. As a result, scientific interest in the field of explainable artificial intelligence (XAI) is gradually emerging. This study aims to review diverse XAI approaches currently exploited in medical imaging. We identify the concepts of the methods, introduce studies applying them to imaging modalities such as computational tomography (CT), magnetic resonance imaging (MRI), and endoscopy, and lastly discuss limitations and challenges faced by XAI for future studies.

• 산업경영시스템학회지
• /
• 제44권4호
• /
• pp.1-11
• /
• 2021

Recently, the importance of preventive maintenance has been emerging since failures in a complex system are automatically detected due to the development of artificial intelligence techniques and sensor technology. Therefore, prognostic and health management (PHM) is being actively studied, and prediction of the remaining useful life (RUL) of the system is being one of the most important tasks. A lot of researches has been conducted to predict the RUL. Deep learning models have been developed to improve prediction performance, but studies on identifying the importance of features are not carried out. It is very meaningful to extract and interpret features that affect failures while improving the predictive accuracy of RUL is important. In this paper, a total of six popular deep learning models were employed to predict the RUL, and identified important variables for each model through SHAP (Shapley Additive explanations) that one of the explainable artificial intelligence (XAI). Moreover, the fluctuations and trends of prediction performance according to the number of variables were identified. This paper can suggest the possibility of explainability of various deep learning models, and the application of XAI can be demonstrated. Also, through this proposed method, it is expected that the possibility of utilizing SHAP as a feature selection method.

• 정보처리학회논문지:소프트웨어 및 데이터공학
• /
• 제11권8호
• /
• pp.339-346
• /
• 2022

최근 화석연료의 무분별한 사용으로 인한 자원고갈 문제 및 기후변화 문제 등이 심각해짐에 따라 화석연료를 대체할 수 있는 신재생에너지에 대한 관심이 증가하고 있다. 특히 신재생에너지 중 태양광 에너지는 다른 신재생에너지원에 비해 고갈될 염려가 적고, 공간적인 제약이 크지 않아 전국적으로 수요가 증가하고 있다. 태양광 발전 시스템에서 생산된 전력을 효율적으로 사용하기 위해서는 보다 정확한 태양광 발전량 예측 모델이 필요하다. 이를 위하여 다양한 기계학습 및 심층학습 기반의 태양광 발전량 예측 모델이 제안되었지만, 심층학습 기반의 예측 모델은 모델 내부에서 일어나는 의사결정 과정을 해석하기가 어렵다는 단점을 보유하고 있다. 이러한 문제를 해결하기 위하여 설명 가능한 인공지능 기술이 많은 주목을 받고 있다. 설명 가능한 인공지능 기술을 통하여 예측 모델의 결과 도출 과정을 해석할 수 있다면 모델의 신뢰성을 확보할 수 있을 뿐만 아니라 해석된 도출 결과를 바탕으로 모델을 개선하여 성능 향상을 기대할 수도 있다. 이에 본 논문에서는 BiLSTM(Bidirectional Long Short-Term Memory)을 사용하여 모델을 구성하고, 모델에서 어떻게 예측값이 도출되었는지를 SHAP(SHapley Additive exPlanations)을 통하여 설명하는 설명 가능한 태양광 발전량 예측 기법을 제안한다.

• 한국융합학회논문지
• /
• 제13권1호
• /
• pp.51-62
• /
• 2022

교통 혼잡을 해결하기 위한 AI 기반 속도 예측 연구는 활발하게 진행되고 있다. 하지만, 인공지능의 추론 과정을 설명하는 설명 가능한 AI의 중요성이 대두되고 있는 가운데 AI 기반 속도 예측의 결과를 해석하고 원인을 추리하는 연구는 미흡하였다. 따라서 본 논문에서는 '설명 가능 그래프 심층 인공신경망 (GNN)'을 고안하여 속도 예측뿐만 아니라, GNN 모델 입력값의 마스킹 기법에 기반하여 인근 도로 영향력을 정량적으로 분석함으로써 혼잡 등의 상황에 대한 추론 근거를 도출하였다. TOPIS 통행 속도 데이터를 활용하여 서울 시내 혼잡 도로를 기준으로 예측 및 분석 방법론을 적용한 후 영향력 높은 인근 도로의 속도를 가상으로 조절하는 시뮬레이션 통하여 혼잡 도로의 통행 속도가 개선됨을 확인하여 제안한 방법론의 타당성을 입증하였다. 이는 교통 네트워크에 제안한 방법론을 적용하고, 그 추론 결과에 기반한 특정 인근 도로를 제어하여 교통 흐름을 개선할 수 있다는 점에 의미가 있다.

• 한국컴퓨터정보학회:학술대회논문집
• /
• 한국컴퓨터정보학회 2024년도 제69차 동계학술대회논문집 32권1호
• /
• pp.35-36
• /
• 2024

전신 마취 수술 중 저혈압의 발생은 다양한 합병증을 유발하며 이를 사전에 예측하여 대응하는 것은 매우 중요한 일이다. 따라서 본 연구에서는 SHAP 모델을 통해 변수 선택을 진행하고, Transformer 모델을 이용해 저혈압 발생 여부를 예측함으로써 임상적 의사결정을 지원한다. 또한 기존 연구들과는 달리, 수술실에서 수집되는 데이터를 기반으로 하여 높은 범용성을 가진다. 비침습적 혈압 예측에서 RMSE 9.46, MAPE 4.4%를 달성하였고, 저혈압 여부를 예측에서는 저혈압 기준 F1-Score 0.75로 우수한 결과를 얻었다.

• Journal of Information Technology Applications and Management
• /
• 제30권1호
• /
• pp.53-70
• /
• 2023

Artificial intelligence has become familiar with modern society, not the distant future. As artificial intelligence and machine learning developed more highly and became more complicated, it became difficult for people to grasp its structure and the basis for decision-making. It is because machine learning only shows results, not the whole processes. As artificial intelligence developed and became more common, people wanted the explanation which could provide them the trust on artificial intelligence. This study recognized the necessity and importance of explainable artificial intelligence, XAI, and examined the trends of XAI research by analyzing social networks and analyzing topics with IEEE published from 2004, when the concept of artificial intelligence was defined, to 2022. Through social network analysis, the overall pattern of nodes can be found in a large number of documents and the connection between keywords shows the meaning of the relationship structure, and topic modeling can identify more objective topics by extracting keywords from unstructured data and setting topics. Both analysis methods are suitable for trend analysis. As a result of the analysis, it was found that XAI's application is gradually expanding in various fields as well as machine learning and deep learning.