• The Korean Journal of Physiology and Pharmacology
• /
• v.23 no.4
• /
• pp.237-249
• /
• 2019

Confirming the direct link between neural circuit activity and animal behavior has been a principal aim of neuroscience. The genetically encoded calcium indicator (GECI), which binds to calcium ions and emits fluorescence visualizing intracellular calcium concentration, enables detection of in vivo neuronal firing activity. Various GECIs have been developed and can be chosen for diverse purposes. These GECI-based signals can be acquired by several tools including two-photon microscopy and microendoscopy for precise or wide imaging at cellular to synaptic levels. In addition, the images from GECI signals can be analyzed with open source codes including constrained non-negative matrix factorization for endoscopy data (CNMF_E) and miniscope 1-photon-based calcium imaging signal extraction pipeline (MIN1PIPE), and considering parameters of the imaged brain regions (e.g., diameter or shape of soma or the resolution of recorded images), the real-time activity of each cell can be acquired and linked with animal behaviors. As a result, GECI signal analysis can be a powerful tool for revealing the functions of neuronal circuits related to specific behaviors.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2022.05a
• /
• pp.475-475
• /
• 2022

The aim of this study is to identify the significant pollutant sources from the tributaries that are affecting the water quality of the study site, the Geum River and provide a solution to enhance the water quality. Multivariate statistical analysis modles such as cluster analysis, Principal component analysis (PCA) and positive matrix factorization (PMF) were applied to identify and prioritize the major pollutant sources of the two major tributaries, Gab-cheon and Miho-cheon, of the Geum River. PCA identifies three major pollutant sources for Gab-cheon and Miho-cheon, respectively. For Gab-cheon, wastewater treatment plant (WWTP), urban, and agricultural pollutions are identified as major pollutant sources. For Miho-cheon, agricultural, urban, and forest land are identified as major pollutant sources. On the contrary, PMF identifies three pollutant sources in Gab-cheon, same as PCA result and two pollutant sources in Miho-cheon. Water quality control scenarios are formulated and improvement of water quality in the river locations are simulated and analyzed with the Environmental Fluid Dynamic Code (EFDC) model. Scenario results were evaluated using a water quality index developed by Canadian Council of Ministers of the Environment. PCA and PMF appears to be effective to identify water pollution sources for the Geum river and also its tributaries in detail and thus can be used for the development of water quality improvement alternative of the above water bodies.

• Annual Conference of KIPS
• /
• 2022.05a
• /
• pp.594-597
• /
• 2022

Recommendation systems are a prominent approach for users to make informed automated judgments. In terms of movie recommendation systems, there are two methods used; Collaborative filtering, which is based on user similarities; and Content-based filtering which takes into account specific user's activity. However, there are still issues with these two existing methods, and to address those, a combination of collaborative and content-based filtering is employed to produce a more effective system. In addition, various similarity methodologies are used to identify parallels among users. This paper focuses on a survey of the various tactics and methods to find solutions based on the problems of the current recommendation system.

• The Korean Journal of Applied Statistics
• /
• v.37 no.3
• /
• pp.311-322
• /
• 2024

In this study, we conduct an exploratory analysis of the MovieLens data, which is frequently used in many recommender system researches, to examine the detailed characteristics of the data. Also, we seek alternatives to improve the well-known neural collaborative filtering (NCF) method. NCF improved matrix factorization method by using deep neural networks in recommender systems. We devise, generalized NCF (G-NCF), a variant of NCF and test the performances. The G-NCF we propose shows superior characteristics on average performance across key evaluation metrics, compared to the NCF, but it also has a slightly larger variance in the evaluation metrics. Evaluation metrics such as MAP and nDCG were considered for comparison.

• Journal of Intelligence and Information Systems
• /
• v.29 no.1
• /
• pp.249-263
• /
• 2023

Collaborative Filtering, a representative recommendation system methodology, consists of two approaches: neighbor methods and latent factor models. Among these, the latent factor model using matrix factorization decomposes the user-item interaction matrix into two lower-dimensional rectangular matrices, predicting the item's rating through the product of these matrices. Due to the factor vectors inferred from rating patterns capturing user and item characteristics, this method is superior in scalability, accuracy, and flexibility compared to neighbor-based methods. However, it has a fundamental drawback: the need to reflect the diversity of preferences of different individuals for items with no ratings. This limitation leads to repetitive and inaccurate recommendations. The Adaptive Deep Latent Factor Model (ADLFM) was developed to address this issue. This model adaptively learns the preferences for each item by using the item description, which provides a detailed summary and explanation of the item. ADLFM takes in item description as input, calculates latent vectors of the user and item, and presents a method that can reflect personal diversity using an attention score. However, due to the requirement of a dataset that includes item descriptions, the domain that can apply ADLFM is limited, resulting in generalization limitations. This study proposes a Generalized Adaptive Deep Latent Factor Recommendation Model, G-ADLFRM, to improve the limitations of ADLFM. Firstly, we use item ID, commonly used in recommendation systems, as input instead of the item description. Additionally, we apply improved deep learning model structures such as Self-Attention, Multi-head Attention, and Multi-Conv1D. We conducted experiments on various datasets with input and model structure changes. The results showed that when only the input was changed, MAE increased slightly compared to ADLFM due to accompanying information loss, resulting in decreased recommendation performance. However, the average learning speed per epoch significantly improved as the amount of information to be processed decreased. When both the input and the model structure were changed, the best-performing Multi-Conv1d structure showed similar performance to ADLFM, sufficiently counteracting the information loss caused by the input change. We conclude that G-ADLFRM is a new, lightweight, and generalizable model that maintains the performance of the existing ADLFM while enabling fast learning and inference.

• Journal of Korean Society for Atmospheric Environment
• /
• v.25 no.2
• /
• pp.108-121
• /
• 2009

The suspended particulate matters have been collected on membrane filters and glass fiber filters by an 8-stage cascade impactor for 2 years (Sep. 2005${\sim}$Sep. 2007) in Kyung Hee University-Global Campus located on the border of Yongin and Suwon. The 20 chemical species (Al, Mn, Si, Fe, Cu, Pb, Cr, Ni, V, Cd, Ba, $Na^+$, ${NH_4}^+$, $K^+$, $Mg^{2+}$, $Ca^{2+}$, $Cl^-$, ${NO_3}^-$, and ${SO_4}^{2-}$) were analyzed by an ICP-AES and an IC after performing proper pre-treatments of each sample filter. Based on these chemical information, the PMF receptor model was applied to identify the source of ambient size-by-size particulate matters. The receptor modeling is the one of the statistical methods to achieve resonable air pollution management strategies. A total of 10 sources was identified in 9 size-ranges such as long-range transport, secondary aerosol, $NH_{4}NO_{3}$ related source, coal combustion, sea-salt, soil, oil combustion, auto emission, incineration, and biomass burning. Especially, the secondary aerosol source assorted in fine and coarse modes was intensively studied.

• Journal of Environmental Science International
• /
• v.17 no.10
• /
• pp.1155-1167
• /
• 2008

Chlorophyll a (chl a) has been used as an indicator for phytoplankton biomass in pelagic ecosystems due to the relative ease of measurement and selectivity for autotrophs in mixed plankton assemblages. However, the use of chi a as an indicator for phytoplankton biomass is restricted due to its inability to resolve taxonomic differences of phytoplankton and the highly variable relationship of chi a with phytoplankton. Here, we describe the analysis of High-Performance Liquid Chromatography (HPLC) photosynthetic pigment data using CHEMTAX, which is a matrix factorization program that uses chemical taxonomic indices (phytoplankton carotenoids) to quantify the abundance of phytoplankton groups. Compared to direct microscopic counting that can distinguish species within broad groups, the resolution of taxonomic groups by CHEMTAX is generally coarse. It can only distinguish between diatoms, dinoflagellates, cryptophytes, cyanobacteria, chlorophytes, prasinophytes, and haptophytes. However, CHEMTAX analysis is much faster and less expensive than microscopic counting methods. HPLC pigment observations were taken in the spring, summer, fall, and winter in$ 2005\sim2006$ within Gamak Bay, South Korea. CHEMTAX results revealed that diatoms were the dominant taxonomic group in Gamak Bay. In inner Gamak Bay, the ratio between diatoms and cryptophytes was $75\sim80%$, and the ratio between dinoflagellates and cryptophytes was $10\sim15%$. In outer Gamak Bay, the ratio between diatoms and cryptophytes was $85\sim90%$, and the ratio between dinflagellates and cryptophytes was only $1\sim5%$. The population structure was seasonal. Relative diatom populations were less in the summer than the winter season.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.4 no.1
• /
• pp.25-44
• /
• 2010

With the increasing popularity of mobile devices, it has become necessary to protect private information and content in these devices. Face recognition has been favored over conventional passwords or security keys, because it can be easily implemented using a built-in camera, while providing user convenience. However, because mobile devices can be used both indoors and outdoors, there can be many illumination changes, which can reduce the accuracy of face recognition. Therefore, we propose a new face recognition method on a mobile device robust to illumination variations. This research makes the following four original contributions. First, we compared the performance of face recognition with illumination variations on mobile devices for several illumination normalization procedures suitable for mobile devices with low processing power. These include the Retinex filter, histogram equalization and histogram stretching. Second, we compared the performance for global and local methods of face recognition such as PCA (Principal Component Analysis), LNMF (Local Non-negative Matrix Factorization) and LBP (Local Binary Pattern) using an integer-based kernel suitable for mobile devices having low processing power. Third, the characteristics of each method according to the illumination va iations are analyzed. Fourth, we use two matching scores for several methods of illumination normalization, Retinex and histogram stretching, which show the best and $2^{nd}$ best performances, respectively. These are used as the inputs of an SVM (Support Vector Machine) classifier, which can increase the accuracy of face recognition. Experimental results with two databases (data collected by a mobile device and the AR database) showed that the accuracy of face recognition achieved by the proposed method was superior to that of other methods.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.19 no.12
• /
• pp.2844-2851
• /
• 2015

How to recover from the multiple 2D images into 3D object has been widely studied in the field of computer vision. In order to improve the accuracy of the recovered 3D shape, it is more important that noise must be minimized and the number of image frames must be guaranteed. However, potential noise is implied when tracking feature points. And the number of image frames which is consisted of an observation matrix usually decrease because of tracking failure, occlusions, or low image resolution, and so on. Therefore, it is obviously essential that the number of image frames must be secured by recovering the missing feature points under noise. Thus, we propose the analytic approach which can control directly the error distance and orientation of missing feature point by the geometrical properties under noise distribution. The superiority of proposed method is demonstrated through experimental results for synthetic and real object.

• Asian Journal of Atmospheric Environment
• /
• v.5 no.2
• /
• pp.86-96
• /
• 2011

The positive matrix factorization (PMF2) and multilinear engine (ME2) models have been shown to be powerful environmental analysis techniques and have been successfully applied to the assessment of ambient particulate matter (PM) source contributions. Because these models are difficult to apply practically, the US EPA developed a more user-friendly version of the PMF. The initial version of the EPA PMF model does not provide any rotational capabilities; for this reason, the model was upgraded to include rotational functions in the EPA PMF ver. 2.0. In this study, PMF and EPA PMF modeling identified ten particulate matter sources including secondary sulfate I, vehicle gasoline, secondary sulfate II, secondary nitrate, secondary sulfate III, incinerators, aged sea salt, airborne soil particles, oil combustion, and diesel emissions. All of the source profiles determined by the two models showed excellent agreement. The calculated average concentrations of $PM_{2.5}$ were consistent between the PMF2 and EPA PMF ($17.94{\pm}0.30{\mu}g/m^3$ and $17.94{\pm}0.30\;{\mu}g/m^3$, respectively). Also, each set of estimated source contributions of the PMF2 and EPA PMF showed good agreement. The results from the new EPA PMF version applying rotational functions were consistent with those of PMF2. Therefore, the updated version of EPA PMF with rotational capabilities will provide more reasonable solutions compared with those of PMF2 and can be more widely applied to air quality management.