• Journal of Korea Multimedia Society
• /
• v.25 no.1
• /
• pp.39-51
• /
• 2022

Image-based gender classification and age estimation of human are classic problems in computer vision. Most of researches in this field focus just only one task of either gender classification or age estimation and most of the reported methods for each task focus on accuracy performance and are not computationally light. Thus, running both tasks together simultaneously on low cost mobile or embedded systems with limited cpu processing speed and memory capacity are practically prohibited. In this paper, we propose a novel light-weight gender classification and age estimation method based on ensemble multitasking deep learning with light-weight processing neural network architecture, which processes both gender classification and age estimation simultaneously and in real-time even for embedded systems. Through experiments over various well-known datasets, it is shown that the proposed method performs comparably to the state-of-the-art gender classification and/or age estimation methods with respect to accuracy and runs fast enough (average 14fps) on a Jestson Nano embedded board.

• Journal of Multimedia Information System
• /
• v.9 no.2
• /
• pp.93-102
• /
• 2022

Dependency parsing is a decision problem of the syntactic relation between words in a sentence. Recently, deep learning models are used for dependency parsing based on the word representations in a continuous vector space. However, it causes a mislabeled tagging problem for the proper nouns that rarely appear in the training corpus because it is difficult to express out-of-vocabulary (OOV) words in a continuous vector space. To solve the OOV problem in dependency parsing, we explored the proper noun embedding method according to the embedding unit. Before representing words in a continuous vector space, we replace the proper nouns with a special token and train them for the contextual features by using the multi-layer bidirectional LSTM. Two models of the syllable-based and morpheme-based unit are proposed for proper noun embedding and the performance of the dependency parsing is more improved in the ensemble model than each syllable and morpheme embedding model. The experimental results showed that our ensemble model improved 1.69%p in UAS and 2.17%p in LAS than the same arc-eager approach-based Malt parser.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2022.11a
• /
• pp.818-820
• /
• 2022

The keyword 'healing' is essential to the competitive society and culture of Koreans. In addition, as the time at home increases due to COVID-19, the demand for indoor healing services has increased. Therefore, this thesis analyzes the user's facial expression so that people can receive various 'customized' healing services indoors, and based on this, provides lighting, ASMR, video recommendation service, and facial expression recording service.The user's expression was analyzed by applying the ensemble algorithm to the expression prediction results of various CNN models after extracting only the face through object detection from the image taken by the user.

• Journal of the Korea Society of Computer and Information
• /
• v.29 no.1
• /
• pp.61-68
• /
• 2024

Despite the continuous efforts to mitigate pedestrian accidents at crosswalks, the problem persist. Vulnerable groups, including the elderly and disabled individuals are at a risk of being involved in traffic incidents. This paper proposes the implementation of object detection algorithm using the YOLO v5 model specifically for pedestrians using assistive devices like wheelchairs and crutches. For this research, data was collected and utilized through image crawling, Roboflow, and Mobility Aids datasets, which comprise of wheelchair users, crutch users, and pedestrians. Data augmentation techniques were applied to improve the model's generalization performance. Additionally, ensemble techniques were utilized to mitigate type 2 errors, resulting in 96% recall rate. This demonstrates that employing ensemble methods with a single YOLO model to target transportation-disadvantaged individuals can yield accurate detection performance without overlooking crucial objects.

• Journal of Korean Society of Industrial and Systems Engineering
• /
• v.44 no.4
• /
• pp.227-233
• /
• 2021

Predictive maintenance has been one of important applications of data science technology that creates a predictive model by collecting numerous data related to management targeted equipment. It does not predict equipment failure with just one or two signs, but quantifies and models numerous symptoms and historical data of actual failure. Statistical methods were used a lot in the past as this predictive maintenance method, but recently, many machine learning-based methods have been proposed. Such proposed machine learning-based methods are preferable in that they show more accurate prediction performance. However, with the exception of some learning models such as decision tree-based models, it is very difficult to explicitly know the structure of learning models (Black-Box Model) and to explain to what extent certain attributes (features or variables) of the learning model affected the prediction results. To overcome this problem, a recently proposed study is an explainable artificial intelligence (AI). It is a methodology that makes it easy for users to understand and trust the results of machine learning-based learning models. In this paper, we propose an explainable AI method to further enhance the explanatory power of the existing learning model by targeting the previously proposedpredictive model [5] that learned data from a core facility (Hyper Compressor) of a domestic chemical plant that produces polyethylene. The ensemble prediction model, which is a black box model, wasconverted to a white box model using the Explainable AI. The proposed methodology explains the direction of control for the major features in the failure prediction results through the Explainable AI. Through this methodology, it is possible to flexibly replace the timing of maintenance of the machine and supply and demand of parts, and to improve the efficiency of the facility operation through proper pre-control.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2022.05a
• /
• pp.579-582
• /
• 2022

청소년의 성격유형을 분석할 때 소셜미디어 데이터를 활용하여 텍스트 처리로 분석하는 연구는 많이 알려져 있다. 그러나 이미지를 사용하여 성격유형을 분석한 연구는 미비하다. 본 연구는 청소년의 발테그 그림검사로 표현된 이미지를 데이터로 사용하고, CNN을 활용하여 MMTIC의 16가지 청소년의 성격유형을 예측한다. 연구 대상은 중학교 재학생을 대상으로 한다. MMTIC에서 U-band를 제외한 340명의 학생으로 2012년 4월부터 2013년 3월까지 조사하였다. 연구 결과 CNN을 사용하였을 때 21.6% 예측율을 보였으며, CNN Ensemble을 적용하였을 때 23.1%로 2.5%가 증가한다.

• KIPS Transactions on Software and Data Engineering
• /
• v.10 no.2
• /
• pp.57-64
• /
• 2021

This study presents a method for realtime fuel consumption prediction using real data collected from OBDII. With the advent of the era of self-driving cars, electronic control units(ECU) are getting more complex, and various studies are being attempted to extract and analyze more accurate data from vehicles. But since ECU is getting more complex, it is getting harder to get the data from ECU. To solve this problem, the firmware was developed for acquiring accurate vehicle data in this study, which extracted 53,580 actual driving data sets from vehicles from January to February 2019. Using these data, the ensemble stacking technique was used to increase the accuracy of the realtime fuel consumption prediction model. In this study, Ridge, Lasso, XGBoost, and LightGBM were used as base models, and Ridge was used for meta model, and the predicted performance was MAE 0.011, RMSE 0.017.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2021.10a
• /
• pp.455-457
• /
• 2021

App is used on mobile devices such as smartphones and also has malicious code, which can be divided into normal and malicious depending on the presence or absence of hacking codes. Because there are many kind of malware, it is difficult to detect directly, we propose a method to detect malicious app using AI. Most of the existing methods are to detect malicious app by extracting features from malicious app. However, the number of types have increased exponentially, making it impossible to detect malicious code. Therefore, we would like to propose two more methods besides detecting malicious app by extracting features from most existing malicious app. The first method is to learn normal app to extract normal's features, as opposed to the existing method of learning malicious app and find abnormalities (malicious app). The second one is an 'ensemble technique' that combines the existing method with the first proposal. These two methods need to be studied so that they can be used in future mobile environment.

• The Journal of the Convergence on Culture Technology
• /
• v.9 no.3
• /
• pp.763-768
• /
• 2023

In this paper, the existing statistical method (ARIMA) and machine learning method (Informer) were developed and compared to predict the distribution volume of pharmaceuticals. It was found that a machine learning-based model is advantageous for daily data prediction, and it is effective to use ARIMA for monthly prediction and switch to Informer as the data increases. The prediction error rate (RMSE) was reduced by 26.6% compared to the previous method, and the prediction accuracy was improved by 13%, resulting in a result of 86.2%. Through this thesis, we find that there is an advantage of obtaining the best results by ensembleing statistical methods and machine learning methods. In addition, machine learning-based AI models can derive the best results through deep learning operations even in irregular situations, and after commercialization, performance is expected to improve as the amount of data increases.

• Smart Structures and Systems
• /
• v.19 no.2
• /
• pp.137-149
• /
• 2017

Under the interference of the temperature effect, the alternation of cable force due to damages of a cable-stayed bridge could be difficult to distinguish. Considering the convenience and applicability in engineering practice, simple air or cable temperature measurements are adopted in the current study for the exclusion of temperature effect from the variation of cable force. Using the data collected from Ai-Lan Bridge located in central Taiwan, this work applies the ensemble empirical mode decomposition to process the time histories of cable force, air temperature, and cable temperature. It is evidently observed that the cable force and both types of temperature can all be categorized as the daily variation, long-term variation, and high-frequency noise in the order of decreasing weight. Moreover, the correlation analysis conducted for the decomposed variations of all these three quantities undoubtedly indicates that the daily and long-term variations with different time shifts have to be distinguished for accurately evaluating the temperature effect on the variation of cable force. Finally, consistent results in reducing the range of cable force variation after the elimination of temperature effect confirm the validity and stability of the developed method.