• 한국멀티미디어학회논문지
• /
• 제23권11호
• /
• pp.1349-1360
• /
• 2020

In recent years, Convolutional Neural Networks (CNNs) have been successfully implemented in different tasks of computer vision. Since CNN models are the representatives of supervised learning algorithms, they demand large amount of data in order to train the classifiers. Thus, obtaining data with correct labels is imperative to attain the state-of-the-art performance of the CNN models. However, labelling datasets is quite tedious and expensive process, therefore real-life datasets often exhibit incorrect labels. Although the issue of poorly labelled datasets has been studied before, we have noticed that the methods are very complex and hard to reproduce. Therefore, in this research work, we propose Deep CleanNet - a considerably simple system that achieves competitive results when compared to the existing methods. We use K-means clustering algorithm for selecting data with correct labels and train the new dataset using a deep CNN model. The technique achieves competitive results in both training and validation stages. We conducted experiments using MNIST database of handwritten digits with 50% corrupted labels and achieved up to 10 and 20% increase in training and validation sets accuracy scores, respectively.

• 한국음향학회지
• /
• 제35권4호
• /
• pp.303-309
• /
• 2016

본 논문은 자기부호화기를 이용한 음향신호 분리방법을 제안한다. 사용된 복층구조 신경망 자기부호화기는 입력 신호의 효율적인 표현방법을 자동으로 학습하며, 유사한 특징을 가지고 있는 요소신호들을 군집함으로써 다른 특징의 신호들을 분리할 수 있다. 시간영역과 주파수영역의 변이특성을 추출하기 위하여 단구간푸리에변환(Short-Time Fourier Transform, STFT)을 수행하였으며, 정해진 크기의 사각형 창을 모든 가능한 위치에 적용하여 얻은 단구간 주파수 스펙트럼을 자기부호화기의 입력으로 사용하였다. 자기부호화기의 부호노드들의 값을 이용하여 유사한 스펙트럼 창들을 군집하고, 이를 이용하여 원래의 음원들로 분리해 낼 수 있었다. 분리된 원음들은 원래의 입력신호의 특징을 확실히 나타내었으며, 기존의 비음수 행렬분해(Non-negative Matrix Factorization, NMF) 결과와 주파수 스펙트럼 비교를 통해 그 유효성을 보일 수 있었다.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제18권2호
• /
• pp.311-326
• /
• 2024

The rapid development of neural network technology promotes the neural network model driven by big data to overcome the texture effect of complex objects. Due to the limitations in complex scenes, it is necessary to establish custom template matching and apply it to the research of many fields of computational vision technology. The dependence on high-quality small label sample database data is not very strong, and the machine learning system of deep feature connection to complete the task of texture effect inference and speculation is relatively poor. The style transfer algorithm based on neural network collects and preserves the data of patterns, extracts and modernizes their features. Through the algorithm model, it is easier to present the texture color of patterns and display them digitally. In this paper, according to the texture effect reasoning of custom template matching, the 3D visualization of the target is transformed into a 3D model. The high similarity between the scene to be inferred and the user-defined template is calculated by the user-defined template of the multi-dimensional external feature label. The convolutional neural network is adopted to optimize the external area of the object to improve the sampling quality and computational performance of the sample pyramid structure. The results indicate that the proposed algorithm can accurately capture the significant target, achieve more ablation noise, and improve the visualization results. The proposed deep convolutional neural network optimization algorithm has good rapidity, data accuracy and robustness. The proposed algorithm can adapt to the calculation of more task scenes, display the redundant vision-related information of image conversion, enhance the powerful computing power, and further improve the computational efficiency and accuracy of convolutional networks, which has a high research significance for the study of image information conversion.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제16권12호
• /
• pp.3836-3854
• /
• 2022

The student group division is benefit for universities to do the student management based on the group profile. With the widespread use of student smart cards on campus, especially where students living in campus residence halls, students' daily activities on campus are recorded with information such as smart card swiping time and location. Therefore, it is feasible to depict the students with the daily activity data and accordingly group students based on objective measuring from their campus behavior with some regular student attributions collected in the management system. However, it is challenge in feature representation due to diverse forms of the student data. To effectively and comprehensively represent students' behaviors for further student group division, we proposed to adopt activity data from student smart cards and student attributes as input data with taking account of activity and attribution relationship types from different perspective. Specially, we propose a novel student group division method based on a multi-view student attribute heterogeneous information network (MSA-HIN). The network nodes in our proposed MSA-HIN represent students with their multi-dimensional attribute information. Meanwhile, the edges are constructed to characterize student different relationships, such as co-major, co-occurrence, and co-borrowing books. Based on the MSA-HIN, embedded representations of students are learned and a deep graph cluster algorithm is applied to divide students into groups. Comparative experiments have been done on a real-life campus dataset collected from a university. The experimental results demonstrate that our method can effectively reveal the variability of student attributes and relationships and accordingly achieves the best clustering results for group division.

• 스마트미디어저널
• /
• 제9권1호
• /
• pp.23-29
• /
• 2020

This paper presents an approach for dynamic hand gesture recognition by using algorithm based on 3D Convolutional Neural Network (3D_CNN), which is later extended to 3D Residual Networks (3D_ResNet), and the neural network based key frame selection. Typically, 3D deep neural network is used to classify gestures from the input of image frames, randomly sampled from a video data. In this work, to improve the classification performance, we employ key frames which represent the overall video, as the input of the classification network. The key frames are extracted by SegNet instead of conventional clustering algorithms for video summarization (VSUMM) which require heavy computation. By using a deep neural network, key frame selection can be performed in a real-time system. Experiments are conducted using 3D convolutional kernels such as 3D_CNN, Inflated 3D_CNN (I3D) and 3D_ResNet for gesture classification. Our algorithm achieved up to 97.8% of classification accuracy on the Cambridge gesture dataset. The experimental results show that the proposed approach is efficient and outperforms existing methods.

• Journal of information and communication convergence engineering
• /
• 제21권3호
• /
• pp.198-207
• /
• 2023

Responding to changes in artificial intelligence models and the data environment is crucial for increasing data-learning accuracy and inference stability of industrial applications. A learning model that is overfitted to specific training data leads to poor learning performance and a deterioration in flexibility. Therefore, an early stopping technique is used to stop learning at an appropriate time. However, this technique does not consider the homogeneity and independence of the data collected by heterogeneous nodes in a differential network environment, thus resulting in low learning accuracy and degradation of system performance. In this study, the generalization performance of neural networks is maximized, whereas the effect of the homogeneity of datasets is minimized by achieving an accuracy of 99.7%. This corresponds to a decrease in delay time by a factor of 2.33 and improvement in performance by a factor of 2.5 compared with the conventional method.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제12권1호
• /
• pp.413-435
• /
• 2018

Although the accuracy of handwritten character recognition based on deep networks has been shown to be superior to that of the traditional method, the use of an overly deep network significantly increases time consumption during parameter training. For this reason, this paper took the training time and recognition accuracy into consideration and proposed a novel handwritten character recognition algorithm with newly designed network structure, which is based on an extended nonlinear kernel residual network. This network is a non-extremely deep network, and its main design is as follows:(1) Design of an unsupervised apriori algorithm for intra-class clustering, making the subsequent network training more pertinent; (2) presentation of an intermediate convolution model with a pre-processed width level of 2;(3) presentation of a composite residual structure that designs a multi-level quick link; and (4) addition of a Dropout layer after the parameter optimization. The algorithm shows superior results on MNIST and SVHN dataset, which are two character benchmark recognition datasets, and achieves better recognition accuracy and higher recognition efficiency than other deep structures with the same number of layers.

• 방송공학회논문지
• /
• 제27권6호
• /
• pp.923-935
• /
• 2022

본 논문에서는 스마트팜 시스템에서 재배 중인 식물 잎의 질병을 검출하고, 질병 유형을 분류하는 방법을 제안한다. 영상으로부터식물 잎의 컬러 정보와 질병 유형의 형태 정보를 다층 퍼셉트론(MLP) 모델을 이용하여 학습한다. 1단계에서는 입력된 영상의 컬러분포를 분석하여 질병 존재 여부를 판단한다. 1단계의 질병 존재 가능성이 높은 영상에 대하여 2단계에서는 Mean shift clustering을 이용하여 작은 영역으로 분할하고, 각 분할된 영역 단위로 컬러 정보를 추출하여 제안한 Color Network에 의하여 질병 여부를 판별한다. 컬러 분할된 영역이 Color Network에 의하여 질병으로 판별되면, 3단계에서는 그 영역의 형태 정보를 추출하여 제안한 Shape Network를 이용하여 질병의 유형을 분류한다. 사과나무 잎과 서양 양상추(Iceberg)에서 발생하는 두 가지 대분류 유형의 질병에 대하여, 제안한 기법은 작은 영역 단위로는 92.3%의 잎 질병 검출률을 보였으며, 보통 2개 이상의 질병 영역이 존재하는 한 장의 영상 단위로는 99.3% 이상의 검출률을 보였다. 본 논문에서 제안한 방법은 스마트팜 환경에서 잎 식물의 질병 여부를 조기에 발견할 수 있으며, 대상 식물에 따른 추가 학습 없이 다양한 식물과 질병 유형으로 확대 적용이 가능하다.

• International Journal of Computer Science & Network Security
• /
• 제22권9호
• /
• pp.23-30
• /
• 2022

Glaucoma is a chronic neuropathy that affects the optic nerve which can lead to blindness. The detection and prediction of glaucoma become possible using deep neural networks. However, the detection performance relies on the availability of a large number of data. Therefore, we propose different frameworks, including a hybrid of a generative adversarial network and a convolutional neural network to automate and increase the performance of glaucoma detection. The proposed frameworks are evaluated using five public glaucoma datasets. The framework which uses a Deconvolutional Generative Adversarial Network (DCGAN) and a DenseNet pre-trained model achieves 99.6%, 99.08%, 99.4%, 98.69%, and 92.95% of classification accuracy on RIMONE, Drishti-GS, ACRIMA, ORIGA-light, and HRF datasets respectively. Based on the experimental results and evaluation, the proposed framework closely competes with the state-of-the-art methods using the five public glaucoma datasets without requiring any manually preprocessing step.

• 한국정보통신학회:학술대회논문집
• /
• 한국정보통신학회 2021년도 추계학술대회
• /
• pp.297-299
• /
• 2021

치밀오일 미래 생산성 예측은 잔류오일 회수량 및 저류층 거동 분석을 위해 중요한 작업이다. 일반적으로 석유공학적 관점에서 감퇴곡선법을 이용하여 생산성 예측이 이루어지는데, 최근에는 데이터기반의 머신러닝 기법을 이용한 연구도 수행되고 있다. 본 연구에서는 딥러닝 기반 순환신경망과 LSTM, GRU 알고리즘을 이용하여 미래 생산량 예측을 위한 효과적인 모델을 제안하고자 한다. 입력변수로는 치밀오일 생산 시 산출되는 오일, 가스, 물과 이와 더불어 다양한 군집분석을 통해 산출된 표준곡선이 주요 매개변수이고, 출력변수는 월별 오일 생산량이다. 기존의 경험적 모델인 감퇴곡선법과 순환신경망 모델들을 비교하였으며, 모델의 예측성능을 향상시키기 위해 하이퍼파라미터 튜닝을 통해 최적 모델을 도출하였다.