• Convergence Security Journal
• /
• v.23 no.4
• /
• pp.61-70
• /
• 2023

Clustering is an unsupervised learning method that involves grouping data based on features such as distance metrics, using data without known labels or ground truth values. This method has the advantage of being applicable to various types of data, including images, text, and audio, without the need for labeling. Traditional clustering techniques involve applying dimensionality reduction methods or extracting specific features to perform clustering. However, with the advancement of deep learning models, research on deep clustering techniques using techniques such as autoencoders and generative adversarial networks, which represent input data as latent vectors, has emerged. In this study, we propose a deep clustering technique based on deep learning. In this approach, we use an autoencoder to transform the input data into latent vectors, and then construct a vector space according to the cluster structure and perform k-means clustering. We conducted experiments using the MNIST and Fashion-MNIST datasets in the PyTorch machine learning library as the experimental environment. The model used is a convolutional neural network-based autoencoder model. The experimental results show an accuracy of 89.42% for MNIST and 56.64% for Fashion-MNIST when k is set to 10.

• Nuclear Engineering and Technology
• /
• v.56 no.6
• /
• pp.2220-2238
• /
• 2024

As a typical active equipment, pump machinery is widely used in nuclear power plants. Although the mechanism of pump machinery in nuclear power plants is similar to that of conventional pumps, the safety and reliability requirements of nuclear pumps are higher in complex operating environments. Once there is significant performance degradation or failure, it may cause huge security risks and economic losses. There are many pumps mechanical parameters, and it is very important to explore the correlation between multi-dimensional variables and condition. Therefore, a condition monitoring model based on Deep Denoising Autoencoder (DDAE) is constructed in this paper. This model not only ensures low false positive rate, but also realizes early abnormal monitoring and location. In order to alleviate the influence of parameter time-varying effect on the model in long-term monitoring, this paper combined equidistant sampling strategy and DDAE model to enhance the monitoring efficiency. By using the simulation data of reactor coolant pump and the actual centrifugal pump data, the monitoring and positioning capabilities of the proposed scheme under normal and abnormal conditions were verified. This paper has important reference significance for improving the intelligent operation and maintenance efficiency of nuclear power plants.

• Annals of Hepato-Biliary-Pancreatic Surgery
• /
• v.26 no.4
• /
• pp.375-385
• /
• 2022

Backgrounds/Aims: Prehabilitation aims for preoperative optimisation to reduce postoperative complications. However, there is a paucity of data on its use in patients undergoing pancreaticoduodenectomy (PD). Thus, this study aims to evaluate the outcomes of a home-based outpatient prehabilitation program (PP) versus no-PP in patients undergoing PD. Methods: This retrospective cohort study compared patients who underwent PP versus no-PP before elective PD from January 2016 to December 2020. Inclusion criteria for PP were < 65 years or 65-74 years with FRAIL score < 3. No-PP included dietician, case manager and anesthesia review. PP included additional physiotherapy sessions, caregiver training and interim phone consultation. Univariate and multivariate analysis were used to evaluate length of stay (LOS), morbidity, 30-day readmission, and 90-day mortality. Results: Seventy-one patients (PP: n = 50 [70.4%]; no-PP: n = 21 [29.6%]) were included in this study. Median age was 65 years (interquartile range [IQR]: 58-72 years). Majority (n = 58 [81.7%]) of patients underwent open surgery. Ductal adenocarcinoma was the most common histology (49.3%). Patient demographics were comparable between both groups. Overall median LOS was 11.0 days (IQR: 8.0-17.0 days). Compared to no-PP, PP was not independently associated with reduced intra-abdominal collections (odds ratio [OR]: 0.43; 95% confidence interval [CI]: 0.03-6.11, p = 0.532), major morbidity (OR: 1.31; 95% CI: 0.09-19.47; p = 0.845) or 30-day readmission (OR: 3.16; 95% CI: 0.26-38.27; p = 0.365). There was one (1.4%) 30-day mortality. Conclusions: Our outpatient PP with unsupervised exercise regimes did not improve postoperative outcomes following elective PD.

• International Journal of Heart Failure
• /
• v.6 no.1
• /
• pp.11-19
• /
• 2024

The prevalence of heart failure (HF) is increasing, necessitating accurate diagnosis and tailored treatment. The accumulation of clinical information from patients with HF generates big data, which poses challenges for traditional analytical methods. To address this, big data approaches and artificial intelligence (AI) have been developed that can effectively predict future observations and outcomes, enabling precise diagnoses and personalized treatments of patients with HF. Machine learning (ML) is a subfield of AI that allows computers to analyze data, find patterns, and make predictions without explicit instructions. ML can be supervised, unsupervised, or semi-supervised. Deep learning is a branch of ML that uses artificial neural networks with multiple layers to find complex patterns. These AI technologies have shown significant potential in various aspects of HF research, including diagnosis, outcome prediction, classification of HF phenotypes, and optimization of treatment strategies. In addition, integrating multiple data sources, such as electrocardiography, electronic health records, and imaging data, can enhance the diagnostic accuracy of AI algorithms. Currently, wearable devices and remote monitoring aided by AI enable the earlier detection of HF and improved patient care. This review focuses on the rationale behind utilizing AI in HF and explores its various applications.

• Korean Journal of Agricultural and Forest Meteorology
• /
• v.8 no.2
• /
• pp.86-96
• /
• 2006

Ecoclimap-1, a new complete surface parameter global database at a 1-km resolution, was previously presented. It is intended to be used to initialize the soil-vegetation- atmosphere transfer schemes in meteorological and climate models. Surface parameters in the Ecoclimap-1 database are provided in the form of a per-class value by an ecoclimatic base map from a simple merging of land cover and climate maps. The principal objective of this ecoclimatic map is to consider intra-class variability of life cycle that the usual land cover map cannot describe. Although the ecoclimatic map considering land cover and climate is used, the intra-class variability was still too high inside some classes. In this study, a new strategy is defined; the idea is to use the information contained in S10 NDVI SPOT/VEGETATION profiles to split a land cover into more homogeneous sub-classes. This utilizes an intra-class unsupervised sub-clustering methodology instead of simple merging. This study was performed to provide a new ecolimatic map over Northeast Asia in the framework of Ecoclimap-2 global database construction for surface parameters. We used the University of Maryland's 1km Global Land Cover Database (UMD) and a climate map to determine the initial number of clusters for intra-class sub-clustering. An unsupervised classification process using six years of NDVI profiles allows the discrimination of different behavior for each land cover class. We checked the spatial coherence of the classes and, if necessary, carried out an aggregation step of the clusters having a similar NDVI time series profile. From the mapping system, 29 ecosystems resulted for the study area. In terms of climate-related studies, this new ecosystem map may be useful as a base map to construct an Ecoclimap-2 database and to improve the surface climatology quality in the climate model.

• Journal of Intelligence and Information Systems
• /
• v.22 no.2
• /
• pp.127-142
• /
• 2016

Deep learning model is a kind of neural networks that allows multiple hidden layers. There are various deep learning architectures such as convolutional neural networks, deep belief networks and recurrent neural networks. Those have been applied to fields like computer vision, automatic speech recognition, natural language processing, audio recognition and bioinformatics where they have been shown to produce state-of-the-art results on various tasks. Among those architectures, convolutional neural networks and recurrent neural networks are classified as the supervised learning model. And in recent years, those supervised learning models have gained more popularity than unsupervised learning models such as deep belief networks, because supervised learning models have shown fashionable applications in such fields mentioned above. Deep learning models can be trained with backpropagation algorithm. Backpropagation is an abbreviation for "backward propagation of errors" and a common method of training artificial neural networks used in conjunction with an optimization method such as gradient descent. The method calculates the gradient of an error function with respect to all the weights in the network. The gradient is fed to the optimization method which in turn uses it to update the weights, in an attempt to minimize the error function. Convolutional neural networks use a special architecture which is particularly well-adapted to classify images. Using this architecture makes convolutional networks fast to train. This, in turn, helps us train deep, muti-layer networks, which are very good at classifying images. These days, deep convolutional networks are used in most neural networks for image recognition. Convolutional neural networks use three basic ideas: local receptive fields, shared weights, and pooling. By local receptive fields, we mean that each neuron in the first(or any) hidden layer will be connected to a small region of the input(or previous layer's) neurons. Shared weights mean that we're going to use the same weights and bias for each of the local receptive field. This means that all the neurons in the hidden layer detect exactly the same feature, just at different locations in the input image. In addition to the convolutional layers just described, convolutional neural networks also contain pooling layers. Pooling layers are usually used immediately after convolutional layers. What the pooling layers do is to simplify the information in the output from the convolutional layer. Recent convolutional network architectures have 10 to 20 hidden layers and billions of connections between units. Training deep learning networks has taken weeks several years ago, but thanks to progress in GPU and algorithm enhancement, training time has reduced to several hours. Neural networks with time-varying behavior are known as recurrent neural networks or RNNs. A recurrent neural network is a class of artificial neural network where connections between units form a directed cycle. This creates an internal state of the network which allows it to exhibit dynamic temporal behavior. Unlike feedforward neural networks, RNNs can use their internal memory to process arbitrary sequences of inputs. Early RNN models turned out to be very difficult to train, harder even than deep feedforward networks. The reason is the unstable gradient problem such as vanishing gradient and exploding gradient. The gradient can get smaller and smaller as it is propagated back through layers. This makes learning in early layers extremely slow. The problem actually gets worse in RNNs, since gradients aren't just propagated backward through layers, they're propagated backward through time. If the network runs for a long time, that can make the gradient extremely unstable and hard to learn from. It has been possible to incorporate an idea known as long short-term memory units (LSTMs) into RNNs. LSTMs make it much easier to get good results when training RNNs, and many recent papers make use of LSTMs or related ideas.

• Journal of KIISE:Software and Applications
• /
• v.27 no.5
• /
• pp.529-542
• /
• 2000

Recent years, the use of multimedia information is rapidly increasing, and the video media is the most rising one than any others, and this field Integrates all the media into a single data stream. Though the availability of digital video is raised largely, it is very difficult for users to make the effective video access, due to its length and unstructured video format. Thus, the minimal interaction of users and the explicit definition of video structure is a key requirement in the lately developing image and video management systems. This paper defines the terms and hierarchical video structure, and presents the system, which construct the clustering-based video hierarchy, which facilitate users by browsing the summary and do a random access to the video content. Instead of using a single feature and domain-specific thresholds, we use multiple features that have complementary relationship for each other and clustering-based methods that use normalization so as to interact with users minimally. The stage of shot boundary detection extracts multiple features, performs the adaptive filtering process for each features to enhance the performance by eliminating the false factors, and does k-means clustering with two classes. The shot list of a result after the proposed procedure is represented as the video hierarchy by the intelligent unsupervised clustering technique. We experimented the static and the dynamic movie videos that represent characteristics of various video types. In the result of shot boundary detection, we had almost more than 95% good performance, and had also rood result in the video hierarchy.

• Journal of the Korean Association of Geographic Information Studies
• /
• v.15 no.2
• /
• pp.113-125
• /
• 2012

This research is aimed at evaluating the land surface characterization of KOMPSAT-3A middle infrared (MIR) data. Airborne Hyperspectral Scanner (AHS) data, which has MIR bands with high spatial resolution, were used to assess land surface temperature (LST) retrieval and classification accuracy of MIR bands. Firstly, LST values for daytime and nighttime, which were calculated with AHS thermal infrared (TIR) bands, were compared to digital number of AHS MIR bands. The determination coefficient of AHS band 68 (center wavelength $4.64{\mu}m$) was over 0.74, and was higher than other MIR bands. Secondly, The land cover maps were generated by unsupervised classification methods using the AHS MIR bands. Each class of land cover maps for daytime, such as water, trees, green grass, roads, roofs, was distinguished well. But some classes of land cover maps for nighttime, such as trees versus green grass, roads versus roofs, were not separated. The image classification using the difference images between daytime AHS MIR bands and nighttime AHS MIR bands were conducted to enhance the discrimination ability of land surface for AHS MIR imagery. The classification accuracy of the land cover map for zone 1 and zone 2 was 67.5%, 64.3%, respectively. It was improved by 10% compared to land cover map of daytime AHS MIR bands and night AHS MIR bands. Consequently, new algorithm based on land surface characteristics is required for temperature retrieval of high resolution MIR imagery, and the difference images between daytime and nighttime was considered to enhance the ability of land surface characterization using high resolution MIR data.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.26 no.2
• /
• pp.93-98
• /
• 2016

For accurate precipitation forecasts the choice of weather factors and prediction method is very important. Recently, machine learning has been widely used for forecasting precipitation, and artificial neural network, one of machine learning techniques, showed good performance. In this paper, we suggest a new method for forecasting precipitation using DBN, one of deep learning techniques. DBN has an advantage that initial weights are set by unsupervised learning, so this compensates for the defects of artificial neural networks. We used past precipitation, temperature, and the parameters of the sun and moon's motion as features for forecasting precipitation. The dataset consists of observation data which had been measured for 40 years from AWS in Seoul. Experiments were based on 8-fold cross validation. As a result of estimation, we got probabilities of test dataset, so threshold was used for the decision of precipitation. CSI and Bias were used for indicating the precision of precipitation. Our experimental results showed that DBN performed better than MLP.

• Journal of the Korea Society for Simulation
• /
• v.29 no.1
• /
• pp.47-55
• /
• 2020

This paper presents a real-time autonomous computation of shot numbers and aiming points against multiple soft targets on grounds by applying an unsupervised learning, k-mean clustering and Monte carlo simulation. For this computation, a 100 × 200 square meters size of virtual battlefield is created where an augmented enemy infantry platoon unit attacks, defences, and is scatted, and a virtual weapon with a lethal range of 15m is modeled. In order to determine damage types of the enemy unit: no damage, light wound, heavy wound and death, Monte carlo simulation is performed to apply the Carlton damage function for the damage effect of the soft targets. In addition, in order to achieve the damage effectiveness of the enemy units in line with the commander's intention, the optimal shot numbers and aiming point locations are calculated in less than 0.4 seconds by applying the k-mean clustering and repetitive Monte carlo simulation. It is hoped that this study will help to develop a system that reduces the decision time for 'detection-decision-shoot' process in battalion-scaled combat units operating Dronebot combat system.