• Proceedings of the Korean Society of Computer Information Conference
• /
• 2019.01a
• /
• pp.463-464
• /
• 2019

본 논문은 기계학습을 기반으로 아두이노와 시리얼통신을 통한 사용자의 동작인식을 이용해 보다 간단하게 인터넷상의 원하는 동영상을 찾을 수 있는 검색시스템을 제작하고자 하였다. 이 검색시스템은 Python을 기반으로 SVM(Support Vector Machine)을 이용한 패턴 분류를 사용하였으며 이를 통해 사용자의 동작을 입력받아 문자를 예측 할 수 있다. 사용자는 이 검색시스템을 사용하기 위하여 우선 문자에 대한 사용자의 동작입력을 통해 학습 데이터 셋을 만들어야 하며 그것을 SVM을 이용하여 학습 모델과 식별자를 만들고, 만들어진 분류기를 통하여 동작인식을 바탕으로 문자의 결과를 예측 할 수 있다. 최종적으로 사용자의 동작인식을 거쳐 만들어진 문자열을 이용해 인터넷 동영상 사이트인 Youtube를 통해 웹 크롤링하여 문자열과 관련 있는 동영상을 찾아준다.

• Annual Conference of KIPS
• /
• 2020.05a
• /
• pp.451-454
• /
• 2020

딥러닝 기술을 이용한 정치적 성향의 편향성 분류를 위하여 신문 뉴스 기사를 수집하고, 머신러닝을 위한 학습 데이터를 구축하였다. 학습 데이터의 구축은 보수 성향과 진보 성향을 대표하는 6개 언론사의 뉴스에서 정치적 성향을 이진 분류 데이터로 구축하였다. 뉴스 기사의 수집 방법으로 최근 이슈들 중에서 정치적 성향과 밀접하게 관련이 있는 키워드 15개를 선정하고 이에 관한 뉴스 기사들을 수집하였다. 그 결과로 11,584개의 학습 및 실험용 데이터를 구축하였으며, 정치적 편향성 분류를 위한 머신러닝 모델을 설계하였다. 머신러닝 기법으로 학습 및 실험을 위해 형태소 단위의 임베딩을 이용하여 문장 및 문서 임베딩으로 확장하였으며, SVM(Support Vector Machine)을 이용하여 정치적 편향성 분류 실험을 수행한 결과로 75%의 정확도를 달성하였다.

• Journal of Radiation Industry
• /
• v.18 no.3
• /
• pp.173-176
• /
• 2024

In this paper, general characteristics, blood tests, and ultrasound examination results were used to classify the presence of polycystic ovary syndrome (PCOS). The classification algorithms used were SVM (Support Vector Machine) and k-NN (k-Nearest Neighbors). Out of a total of 300 data samples, 210 were used as training data and 90 as test data. The results showed that SVM achieved higher accuracy compared to k-NN, confirming its greater utility in diagnosing the presence of PCOS. Future research is expected to improve classification performance by incorporating various additional indicators and securing more data. Additionally, it is expected to serve as a foundational resource for predicting and classifying other diseases.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.21 no.12
• /
• pp.2298-2304
• /
• 2017

Online signature verification is one of the simple and efficient method of identity verification and has less resistance than other biometric technologies. In training to build a verification model, negative samples are required to build the model, but in most practical applications it is not easy to get negative samples - forgery signatures. In this paper, proposed is a method using someone else's signatures as negative samples. In verification, shape-based features extracted from the time-sequenced signature data are extracted and a support vector machine is used to verify. SVM tries to map a feature vector to a high dimensional space and to draw a linear boundary in the high dimensional space. SVM is one of the best classifiers and has been applied to various applications. Using general handwriting data, i.e., someone else's signatures which have little in common with positive samples improved the verification rate experimentally, which means that signature verification without negative samples is possible.

• Journal of the Korea Convergence Society
• /
• v.8 no.4
• /
• pp.1-8
• /
• 2017

In the modern society, the size of the fashion market is continuously increasing both overseas and domestic. When purchasing a product through e-commerce, the evaluation data for the product created by other consumers has an effect on the consumer's decision to purchase the product. By analysing the consumer's evaluation data on the product the company can reflect consumer's opinion which can leads to positive affect of performance to company. In this paper, we propose a method to construct a model to analyze user's sentiment using word embedding space formed by learning review data of amazon fashion products. Experiments were conducted by learning three SVM classifiers according to the number of positive and negative review data using the formed word embedding space which is formed by learning 5.7 million Amazon review data.. Experimental results showed the highest accuracy of 88.0% when learning SVM classifier using 50,000 positive review data and 50,000 negative review data.

• Journal of Korea Society of Industrial Information Systems
• /
• v.28 no.4
• /
• pp.11-19
• /
• 2023

The paper proposes a smart mirror system that recommends fragrances based on user emotion analysis. This paper combines natural language processing techniques such as embedding techniques (CounterVectorizer and TF-IDF) and machine learning classification models (DecisionTree, SVM, RandomForest, SGD Classifier) to build a model and compares the results. After the comparison, the paper constructs a personal emotion-based fragrance recommendation mirror model based on the SVM and word embedding pipeline-based emotion classifier model with the highest performance. The proposed system implements a personalized fragrance recommendation mirror based on emotion analysis, providing web services using the Flask web framework. This paper uses the Google Speech Cloud API to recognize users' voices and use speech-to-text (STT) to convert voice-transcribed text data. The proposed system provides users with information about weather, humidity, location, quotes, time, and schedule management.

• Journal of Korea Water Resources Association
• /
• v.52 no.12
• /
• pp.1075-1086
• /
• 2019

Recently, as the incidence of climate warming and abnormal climate increases, the forecasting of hydrological factors such as precipitation and river flow is getting more complicated, and the risk of water shortage is also increasing. Therefore, this study aims to develop a model for predicting the amount of water intake in mid-term. To this end, the correlation between water intake and meteorological factors, including temperature and precipitation, was used to select input factors. In addition, the amount of water intake increased with time series and seasonal characteristics were clearly shown. Thus, the preprocessing process was performed using the time series decomposition method, and the support vector machine (SVM) was applied to the residual to develop the river intake prediction model. This model has an error of 4.1% on average, which is higher accuracy than the SVM model without preprocessing. In particular, this model has an advantage in mid-term prediction for one to two months. It is expected that the water intake forecasting model developed in this study is useful to be applied for water allocation computation in the permission of river water use, water quality management, and drought measurement for sustainable and efficient management of water resources.

• The Journal of the Korea Contents Association
• /
• v.12 no.4
• /
• pp.40-51
• /
• 2012

Online discussion bulletin in Korea is not only a specific place where user exchange opinions but also a public sphere through which users discuss and form public opinion. Sometimes, there is a heated debate on a topic and any article becomes a political or sociological issue. In this paper, we propose how to analyze the popularity of articles by collecting the information of articles obtained from two well-known discussion forums such as AGORA and SEOPRISE. And we propose a prediction model for the article popularity by applying the characteristics of subject articles. Our experiment shown that the popularity of 87.52% articles have been saturated within a day after the submission in AGORA, but the popularity of 39% articles is growing after 4 days passed in SEOPRISE. And we observed that there is a low correlation between the period of popularity and the hit count. The steady increase of the hit count of an article does not necessarily imply the final hit count of the article at the saturation point is so high. In this paper, we newly propose a new prediction model called 'baseline'. We evaluated the predictability for popular articles using three models (SVM, similar matching and baseline). Through the results of performance evaluation, we observed that SVM model is the best in F-measure and precision, but baseline is the best in running time.

• Journal of the Institute of Electronics Engineers of Korea CI
• /
• v.49 no.1
• /
• pp.16-22
• /
• 2012

In this paper, we proposed a method of baseline correction for analysis of Raman spectra of platelets from Alzheimer's disease (AD) transgenic mice. Measured Raman spectra include the meaningful information and unnecessary noise which is composed of baseline and additive noise. The Raman spectrum is divided into the local region including several peaks and the spectrum of the region is modeled by curve fitting using Gaussian model. The additive noise is clearly removed from the process of replacing the original spectrum with the fitted model. The baseline correction after interpolating the local minima of the fitted model with linear, piecewise cubic Hermite and cubic spline algorithm. The baseline corrected models extract the feature with principal component analysis (PCA). The classification result of support vector machine (SVM) and maximum $a$ posteriori probability (MAP) using linear interpolation method showed the good performance about overall number of principal components, especially SVM gave the best performance which is about 97.3% true classification average rate in case of piecewise cubic Hermite algorithm and 5 principal components. In addition, it confirmed that the proposed baseline correction method compared with the previous research result could be effectively applied in the analysis of the Raman spectra of platelet.

• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.31 no.1
• /
• pp.83-97
• /
• 2021

Wireless sensors that make up the Wireless Sensor Network generally have extremely limited power and resources. The wireless sensor enters the sleep state at a certain interval to conserve power. The Sleep deflation attack is a deadly attack that consumes power by preventing wireless sensors from entering the sleep state, but there is no clear countermeasure. Thus, in this paper, using clustering-based binary search tree structure, the Sleep deprivation attack detection model is proposed. The model proposed in this paper utilizes one of the characteristics of both attack sensor nodes and normal sensor nodes which were classified using machine learning. The characteristics used for detection were determined using Long Short-Term Memory, Decision Tree, Support Vector Machine, and K-Nearest Neighbor. Thresholds for judging attack sensor nodes were then learned by applying the SVM. The determined features were used in the proposed algorithm to calculate the values for attack detection, and the threshold for determining the calculated values was derived by applying SVM.Through experiments, the detection model proposed showed a detection rate of 94% when 35% of the total sensor nodes were attack sensor nodes and improvement of up to 26% in power retention.