• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제13권11호
• /
• pp.5427-5445
• /
• 2019

While deep neural networks have achieved remarkable performance in representation learning, a huge amount of labeled training data are usually required by supervised deep models such as convolutional neural networks. In this paper, we propose a new representation learning method, namely generative adversarial networks (GAN) based bagging deep convolutional autoencoders (GAN-BDCAE), which can map data to diverse hierarchical representations in an unsupervised fashion. To boost the size of training data, to train deep model and to aggregate diverse learning machines are the three principal avenues towards increasing the capabilities of representation learning of neural networks. We focus on combining those three techniques. To this aim, we adopt GAN for realistic unlabeled sample generation and bagging deep convolutional autoencoders (BDCAE) for robust feature learning. The proposed method improves the discriminative ability of learned feature embedding for solving subsequent pattern recognition problems. We evaluate our approach on three standard benchmarks and demonstrate the superiority of the proposed method compared to traditional unsupervised learning methods.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제12권8호
• /
• pp.3790-3803
• /
• 2018

Content-based image retrieval is an approach used to query images based on their semantics. Semantic based retrieval has its application in all fields including medicine, space, computing etc. Semantically generated binary hash codes can improve content-based image retrieval. These semantic labels / binary hash codes can be generated from unlabeled data using convolutional autoencoders. Proposed approach uses semi-supervised deep hashing with semantic learning and binary code generation by minimizing the objective function. Convolutional autoencoders are basis to extract semantic features due to its property of image generation from low level semantic representations. These representations of images are more effective than simple feature extraction and can preserve better semantic information. Proposed activation and loss functions helped to minimize classification error and produce better hash codes. Most widely used datasets have been used for verification of this approach that outperforms the existing methods.

• 대한원격탐사학회지
• /
• 제33권4호
• /
• pp.423-436
• /
• 2017

This paper presents a deep learning-based road segmentation framework from very high-resolution orthophotos. The proposed method uses Deep Convolutional Autoencoders for end-to-end mapping of orthophotos to road segmentations. In addition, a set of post-processing steps were applied to make the model outputs GIS-ready data that could be useful for various applications. The optimization of the model's parameters is explained which was conducted via grid search method. The model was trained and implemented in Keras, a high-level deep learning framework run on top of Tensorflow. The results show that the proposed model with the best-obtained hyperparameters could segment road objects from orthophotos at an average accuracy of 88.5%. The results of optimization revealed that the best optimization algorithm and activation function for the studied task are Stochastic Gradient Descent (SGD) and Exponential Linear Unit (ELU), respectively. In addition, the best numbers of convolutional filters were found to be 8 for the first and second layers and 128 for the third and fourth layers of the proposed network architecture. Moreover, the analysis on the time complexity of the model showed that the model could be trained in 4 hours and 50 minutes on 1024 high-resolution images of size $106{\times}106pixels$, and segment road objects from similar size and resolution images in around 14 minutes. The results show that the deep learning models such as Convolutional Autoencoders could be a best alternative to traditional machine learning models for road segmentation from aerial photographs.

• 한국멀티미디어학회논문지
• /
• 제23권1호
• /
• pp.1-7
• /
• 2020

Clustering methods essentially take a two-step approach; extracting feature vectors for dimensionality reduction and then employing clustering algorithm on the extracted feature vectors. However, for clustering images, the traditional clustering methods such as stacked auto-encoder based k-means are not effective since they tend to ignore the local information. In this paper, we propose a method first to effectively reduce data dimensionality using convolutional auto-encoder to capture and reflect the local information and then to accurately cluster similar data samples by using a hierarchical clustering approach. The experimental results confirm that the clustering results are improved by using the proposed model in terms of clustering accuracy and normalized mutual information.

• 한국멀티미디어학회논문지
• /
• 제22권3호
• /
• pp.335-348
• /
• 2019

Cell segmentation and counting represent one of the most important tasks required in order to provide an exhaustive understanding of biological images. Conventional features suffer the lack of spatial consistency by causing the joining of the cells and, thus, complicating the cell counting task. We propose, in this work, a cascade of networks that take as inputs different versions of the original image. After constructing a Gaussian pyramid representation of the microscopy data, the inputs of different size and spatial resolution are given to a cascade of deep convolutional autoencoders whose task is to reconstruct the segmentation mask. The coarse masks obtained from the different networks are summed up in order to provide the final mask. The principal and main contribution of this work is to propose a novel method for the cell counting. Unlike the majority of the methods that use the obtained segmentation mask as the prior information for counting, we propose to utilize the hidden latent representations, often called the high-level features, as the inputs of a neural network based regressor. While the segmentation part of our method performs as good as the conventional deep learning methods, the proposed cell counting approach outperforms the state-of-the-art methods.

• International Journal of Computer Science & Network Security
• /
• 제22권5호
• /
• pp.73-78
• /
• 2022

The aim of this paper is to investigate the prevalence of Deep Learning in the literature on Fire & Rescue Service. It is found that deep learning techniques are only beginning to benefit the firefighters. The popular areas where deep learning techniques are making an impact are situational awareness, decision making, mental stress, injuries, well-being of the firefighter such as his sudden fall, inability to move and breathlessness, path planning by the firefighters while getting to an fire scene, wayfinding, tracking firefighters, firefighter physical fitness, employment, prediction of firefighter intervention, firefighter operations such as object recognition in smoky areas, firefighter efficacy, smart firefighting using edge computing, firefighting in teams, and firefighter clothing and safety. The techniques that were found applied in firefighting were Deep learning, Traditional K-Means clustering with engineered time and frequency domain features, Convolutional autoencoders, Long Short-Term Memory (LSTM), Deep Neural Networks, Simulation, VR, ANN, Deep Q Learning, Deep learning based on conditional generative adversarial networks, Decision Trees, Kalman Filters, Computational models, Partial Least Squares, Logistic Regression, Random Forest, Edge computing, C5 Decision Tree, Restricted Boltzmann Machine, Reinforcement Learning, and Recurrent LSTM. The literature review is centered on Firefighters/firemen not involved in wildland fires. The focus was also not on the fire itself. It must also be noted that several deep learning techniques such as CNN were mostly used in fire behavior, fire imaging and identification as well. Those papers that deal with fire behavior were also not part of this literature review.

• 융합보안논문지
• /
• 제23권4호
• /
• pp.61-70
• /
• 2023

클러스터링은 데이터의 정답값(실제값)이 없는 데이터를 기반으로 데이터의 특징벡터의 거리 기반 등으로 군집화를 하는 비지도학습 방법이다. 이 방법은 이미지, 텍스트, 음성 등 다양한 데이터에 대해서 라벨링이 없이 적용할 수 있다는 장점이 있다. 기존 클러스터링을 하기 위해 차원축소 기법을 적용하거나 특정 특징만을 추출하여 군집화하는 방법이 적용되었다. 하지만 딥러닝 기반 모델이 발전하면서 입력 데이터를 잠재 벡터로 표현하는 오토인코더, 생성 적대적 네트워크 등을 통해서 딥 클러스터링의 기술이 연구가 되고 있다. 본 연구에서, 딥러닝 기반의 딥 클러스터링 기법을 제안하였다. 이 방법에서 오토인코더를 이용하여 입력 데이터를 잠재 벡터로 변환하고 이 잠재 벡터를 클러스터 구조에 맞게 벡터 공간을 구성 및 k-평균 클러스터링을 하였다. 실험 환경으로 pytorch 머신러닝 라이브러리를 이용하여 데이터셋으로 MNIST와 Fashion-MNIST을 적용하였다. 모델로는 컨볼루션 신경망 기반인 오토인코더 모델을 사용하였다. 실험결과로 k가 10일 때, MNIST에 대해서 89.42% 정확도를 가졌으며 Fashion-MNIST에 대해서 56.64% 정확도를 가진다.

• 지구물리와물리탐사
• /
• 제23권3호
• /
• pp.117-130
• /
• 2020

지구물리탐사 자료의 잡음은 물리탐사 자료를 왜곡시켜 잘못된 결과 해석을 유도한다. 잡음을 만들어내는 원인으로는 인간의 활동으로 인하며 만들어지는 잡음과 자연 현상 및 기기 소음 등이 있으며 이러한 잡음을 제거하기 위한 다양한 연구들이 진행되고 있다. 하지만, 전통적인 잡음제거 방법들은 요소파 변환이나 필터링 과정에서 개인의 주관과 높은 계산 비용 그리고 많은 시간이 소모된다는 단점이 있으며 이런 문제를 해결하기 위해 영상 전처리 및 잡음제거를 위한 개선된 신경망을 구현하고자 하였다. 이 연구는 인공신경망, 합성곱 신경망, 오토인코더, 잔차 및 파형신경망의 다양한 유형의 신경망과 탄성파, 시간영역 전자탐사, 지표투과레이더 및 자기지전류의 잡음을 분석하고, 훈련 과정에 실제로 이용한 인공 신경망과 제시된 핵심 해결책을 분석 정리하였다. 이러한 분석을 통해 개선된 신경망이 지구물리탐사 자료의 잡음제거에 유용한 기법임을 알 수 있었다.

• 인터넷정보학회논문지
• /
• 제22권1호
• /
• pp.23-30
• /
• 2021

본 논문은 다수의 오토인코더 모델들을 이용한 잡음에 강인한 이미지 분류 시스템을 제안한다. 딥러닝 기술의 발달로 이미지 분류의 정확도는 점점 높아지고 있다. 하지만 입력 이미지가 잡음에 의해서 오염된 경우에는 이미지 분류 성능이 급격히 저하된다. 이미지에 첨가되는 잡음은 이미지의 생성 및 전송 과정에서 필연적으로 발생할 수밖에 없다. 따라서 실제 환경에서 이미지 분류기가 사용되기 위해서는 잡음에 대한 처리 및 대응이 반드시 필요하다. 한편 오토인코더는 입력값과 출력값이 유사하도록 학습되어지는 인공신경망 모델이다. 입력데이터가 학습데이터와 유사하다면 오토인코더의 출력데이터와 입력데이터 사이의 오차는 작을 것이다. 하지만 입력 데이터가 학습데이터와 유사성이 없다면 오토인코더의 출력데이터와 입력데이터 사이의 오차는 클 것이다. 제안하는 시스템은 오토인코더의 입력데이터와 출력데이터 사이의 관계를 이용한다. 제안하는 시스템의 이미지 분류 절차는 2단계로 구성된다. 1단계에서 분류 가능성이 가장 높은 클래스 2개를 선정하고 이들 클래스의 분류 가능성이 서로 유사하면 2단계에서 추가적인 분류 절차를 거친다. 제안하는 시스템의 성능 분석을 위해 가우시안 잡음으로 오염된 MNIST 데이터셋을 대상으로 분류 정확도를 실험하였다. 실험 결과 잡음 환경에서 제안하는 시스템이 CNN(Convolutional Neural Network) 기반의 분류 기법에 비해 높은 정확도를 나타냄을 확인하였다.