• Title/Summary/Keyword: Multiple trees

Search Result 197, Processing Time 0.024 seconds

Research on Optimization Strategies for Random Forest Algorithms in Federated Learning Environments (연합 학습 환경에서의 랜덤 포레스트 알고리즘 최적화 전략 연구)

  • InSeo Song;KangYoon Lee
    • The Journal of Bigdata
    • /
    • v.9 no.1
    • /
    • pp.101-113
    • /
    • 2024
  • Federated learning has garnered attention as an efficient method for training machine learning models in a distributed environment while maintaining data privacy and security. This study proposes a novel FedRFBagging algorithm to optimize the performance of random forest models in such federated learning environments. By dynamically adjusting the trees of local random forest models based on client-specific data characteristics, the proposed approach reduces communication costs and achieves high prediction accuracy even in environments with numerous clients. This method adapts to various data conditions, significantly enhancing model stability and training speed. While random forest models consist of multiple decision trees, transmitting all trees to the server in a federated learning environment results in exponentially increasing communication overhead, making their use impractical. Additionally, differences in data distribution among clients can lead to quality imbalances in the trees. To address this, the FedRFBagging algorithm selects only the highest-performing trees from each client for transmission to the server, which then reselects trees based on impurity values to construct the optimal global model. This reduces communication overhead and maintains high prediction performance across diverse data distributions. Although the global model reflects data from various clients, the data characteristics of each client may differ. To compensate for this, clients further train additional trees on the global model to perform local optimizations tailored to their data. This improves the overall model's prediction accuracy and adapts to changing data distributions. Our study demonstrates that the FedRFBagging algorithm effectively addresses the communication cost and performance issues associated with random forest models in federated learning environments, suggesting its applicability in such settings.

A Study on Vegetation Changes for 11years and Vegetation Structure in the Green Buffer Zone of Sihwa Industrial Complex (시화공단 완충녹지의 11년간(2006~2017년) 식생변화 및 식생구조 연구)

  • Choi, Jin-Woo
    • Journal of the Korean Society of Environmental Restoration Technology
    • /
    • v.21 no.5
    • /
    • pp.81-96
    • /
    • 2018
  • The purpose of this study is to analyze the characteristics of vegetation changes and structures in the buffer green zone of Sihwa Industrial Complex and propose environmental, ecological and multi-functional vegetation management directions. The density of the Pinus thunbergii decreased from $23.1trees/100m^2$ in 2006 to $9.6trees/100m^2$ in 2017 due to the influence of thinning works. Its green volume increased from $0.97m^3/m^2$ to $2.02m^3/m^2$, contributing to the improvement of the atmospheric environment and promotion of biodiversity. The density of deciduous broad-leaved trees at the top maintained at $18{\sim}21trees/100m^2$. In the areas where the understory vegetation was removed, Pueraria lobata spread as an invasive plant and disturbed the forest. In the areas where the understory vegetation was not removed, various native plant species including the Rhus javanica and Celtis sinensis were introduced and contributed to the development of a forest in a multiple structure. There was a concern with the spread of Robinia pseudoacacia which developed into understory vegetation in Pinus thunbergii forest and some canopy forest. The study proposed vegetation management directions for the back, top and front side according to the physical structure of large-scale buffer green zone formed in a long linear form and the usage characteristics of adjacent land.

Effect of Carbon Sequestration and Oxygen Production of Trees on Kangwon National University Campus

  • Hyeong-Uk Ahn;Yun Eui Choi;Sung-Ho Kil;Hyun-Kil Jo
    • Journal of Forest and Environmental Science
    • /
    • v.39 no.3
    • /
    • pp.128-139
    • /
    • 2023
  • Urban forests serve multiple purposes by providing green resting spaces and environmental benefits for city residents. In the old city center, where parks are scarce, the campus of Kangwon National University, Chuncheon, Gangwon-do, South Korea, serves as an urban forest for students, faculty, and citizens. This study aims to quantitatively analyze the environmental functions of green spaces on campus, raising awareness about their importance among campus members. The total carbon storage of campus trees was estimated at 1,653,218 kg, including 1,512,586 kg in forest areas, 131,061 kg in planting spaces around buildings, and 9,571 kg in street spaces. The annual carbon uptake of campus trees was estimated to be 39,391 kg/year, with 30,144 kg/year in forest areas, 8,017 kg/year in planting spaces around buildings, and 1,230 kg/year in horizontal spaces. In addition, annual oxygen production was estimated to be 105,044 kg/year, with 80,385 kg/year in forest areas, 21,378 kg/year in planting spaces around buildings, and 3,281 kg/year in street spaces. Furthermore, we estimated carbon emissions from the use of on-campus facilities to be 4,856,182 kg/year, while oxygen consumption by members was estimated at 53,975 kg/year. However, the campus trees supplied a sufficient amount of oxygen, which was twice the amount required by school members. The carbon uptake amount was approximately 1% of the amount of carbon emissions, resulting in a modest contribution to improving the environmental conditions of the site.

Rule-based System for Loading Multiple Items in Containers for Shipping (제품수송 컨터네이너의 적재를 위한 규칙기반시스템)

  • Park, Ji Hee;Lee, Gun Ho
    • KIPS Transactions on Software and Data Engineering
    • /
    • v.2 no.6
    • /
    • pp.403-412
    • /
    • 2013
  • This study figures out the concepts of container transport, logistical cost and the distribution of a company through studying documents, and to suggest logistical cost reduction approach, focused on the efficiency of transport which occupied the considerable portion of the total logistical cost of the company. We analyze and discuss the container loading of multiple items for multiple places of departure and arrival through a case study on S company in South Korea. We suggest a direction to reduce the logistical cost of the companies, analyzing the conditions of multiple items loading, and rule-based systems including an algorithm which determines container-loading for minimum freight expenses. We use data mining and OLAP tools of MS Analysis Services to produce loading rules for multiple items loading and generate OLAP cube and decision trees to validate the rules.

Approach to diagnosing multiple abnormal events with single-event training data

  • Ji Hyeon Shin;Seung Gyu Cho;Seo Ryong Koo;Seung Jun Lee
    • Nuclear Engineering and Technology
    • /
    • v.56 no.2
    • /
    • pp.558-567
    • /
    • 2024
  • Diagnostic support systems are being researched to assist operators in identifying and responding to abnormal events in a nuclear power plant. Most studies to date have considered single abnormal events only, for which it is relatively straightforward to obtain data to train the deep learning model of the diagnostic support system. However, cases in which multiple abnormal events occur must also be considered, for which obtaining training data becomes difficult due to the large number of combinations of possible abnormal events. This study proposes an approach to maintain diagnostic performance for multiple abnormal events by training a deep learning model with data on single abnormal events only. The proposed approach is applied to an existing algorithm that can perform feature selection and multi-label classification. We choose an extremely randomized trees classifier to select dedicated monitoring parameters for target abnormal events. In diagnosing each event occurrence independently, two-channel convolutional neural networks are employed as sub-models. The algorithm was tested in a case study with various scenarios, including single and multiple abnormal events. Results demonstrated that the proposed approach maintained diagnostic performance for 15 single abnormal events and significantly improved performance for 105 multiple abnormal events compared to the base model.

Indicators for the Quantitative Assessment of Tree Vigor Condition and Its Theoretical Implications : A Case Study of Japanese Flowering-cherry Trees in Urban Park (도시공원에 식재된 왕벚나무 수종을 중심으로 한 수목활력도의 정량평가지표 개발 및 이론적 고찰에 관한 연구)

  • Song, Youngkeun
    • Journal of the Korean Society of Environmental Restoration Technology
    • /
    • v.17 no.4
    • /
    • pp.57-67
    • /
    • 2014
  • The vigor condition of trees is an important indicator for the management of urban forested area. But difficulties in how to assess the tree vigor condition still remain. Previous efforts were limited in the 1) measurement of single indicator rather than using multiple indices, 2) purpose-oriented measurement such as for air-pollution effect or specific pathological symptom, and 3) ordinal-scale evaluations by field crews 4) despite human errors based on his/her experiences or prior knowledge. Therefore, this study attempted to develop a quantitative and objective methodology for assessing tree vigor condition, by measuring multiple modules and building the profile inventory. Furthermore, the possibility and limitations were discussed in terms of schematic frames describing tree vigor condition. The vigor condition of 56 flowering cherry plants in urban park were assessed by in-situ measurements of following eight items; growth of crown(Gc), growth of shoots, individual tree volume(Vol), plant area index, woody area index, leaf area index, leaf chlorophyll content(Lc) and leaf water content(Lw). For validation, these measurements were compared with the ranks of holistic tree vigor condition, which were visually assessed using a 4-point grading scale based on the expert's knowledge. As a result, the measures of each evaluation item successfully highlighted a variety of aspects in tree vigor condition, including the states of both photosynthetic and non-photosynthetic parts. The variation in the results depending on evaluated parts was shown within an individual tree, even though the broad agreement among the results was found. The result of correlation analysis between the tested measurements and 4-point visual assessment, demonstrated that the state of water-stressed foliage of the season (Lw) or the development of plant materials since sapling phase (Vol) could be better viewed from the outer appearance of trees than other symptoms. But only based on the visual assessment, it may be difficult to detect the quality of photosynthesis (Lc) or the recent trend in growth of trees (Gc). To make this methodology simplified for the broad-scale application, the tested eight measurements could be integrated into two components by principal component analysis, which was labelled with 'the amount of plant materials' and 'vigor trend', respectively. In addition, the use of these quantitative and multi-scale indicators underlies the importance of assessing various aspects of tree vigor condition, taking into account the response(s) on different time and spatial scale of pressure(s) shown in each evaluated module. Future study should be advanced for various species at diverse developing stages and environment, and the application to wide areas at a periodic manner.

Multiple Applications of Lime Sulfur for Fruit Thinning of 'Fuji' and 'Hongro' Apple Trees (석회유황합제의 다중처리에 따른 '후지' 및 '홍로' 사과의 적과 효과)

  • Chun, Ik-Jo;Zheng, Wei-Wei;Choi, Cheol;Song, Yang-Yik;Kwang, In-Kye;Hirst, Peter
    • Journal of Bio-Environment Control
    • /
    • v.21 no.4
    • /
    • pp.445-451
    • /
    • 2012
  • The thinning effects of lime sulfur either alone or combined with fish oil on two apple cultivars 'Fuji'/M.9 and 'Hongro'/M.9 were investigated during two consecutive seasons. Either 1% or 2% lime sulfur at full bloom decreased the amounts of both terminal and axillary fruits compared with unsprayed 'Fuji' control trees in the first year. However, 1% or 2% lime sulfur did not decrease terminal fruit set in the second year. Only, triple applications of 1% lime sulfur significantly reduced the number of terminal and axillary fruits in both years. Multiple application of 1% lime sulfur was the most effective in thinning of 'Hongro'. In the first year, double or triple applications of 1% lime sulfur was effective in thinning terminal or axillary fruit, but one time applications of 1% or 2% lime sulfur was not reduced axillary fruit in 'Hongro' apple trees. In the second season, all treatments reduced terminal or axillary fruit in 'Hongro' clusters, except single application of 1% lime sulfur. Fish oil did not always improve the thinning effects of lime sulfur in 'Fuji' or 'Hongro' apple trees. None of treatments caused significant russeting in the harvested fruits. Most of thinning treatments had little effect on 'Fuji' or 'Hongro' fruit quality in terms of fruit weight, firmness, soluble solids concentration or titratable acidity.

A Study on Fault Tree Construction for Track Worker's Risk Assessment (선로 작업자 위험도 예측을 위한 고장수목 구성 연구)

  • Kwak Sang-Log;Wang Jong-Bae;Park Chan-Woo;Cho Yuen-Ok
    • Proceedings of the KSR Conference
    • /
    • 2005.05a
    • /
    • pp.123-126
    • /
    • 2005
  • Recently many accidents have been occurred on track workers, these accidents have strong relationship with increase of train speed, electrification and multiple track portion. As a first step for the safety management, domestic and abroad track worker accidents data are analysed for the risk estimation of track worker. Analysis results shows that contact between track worker and train is the dormant reason. In order to reduce dormant reason fault trees are constructed in this study.

  • PDF

Implementation of Arithmetic Processor Using Multi-Valued Logic (다치 논리를 이용한 연산기 구현)

  • 양대영;김휘진;박진우;송홍복
    • Proceedings of the Korean Institute of Information and Commucation Sciences Conference
    • /
    • 1998.05a
    • /
    • pp.338-341
    • /
    • 1998
  • This paper presents CMOS full adder design method based on carry-propagation-free addition trees and a circuit technique, so called multiple-valued current-nude(MVCM) circuits. The carry-propagation-free addition method uses a redundant digit sets called redundant positive-digit number representations. The carry-propagation-free addition is by three steps, and the adder can be designed directly and efficiently from the algorithm using WVCM circuit, Also Multiplier can be designed by these adder. We demonstrate the effectiveness of the proposed method through simulation(SPICE).

  • PDF

Generation of Pattern Classifiers Based on Linear Nongroup CA

  • Choi, Un-Sook;Cho, Sung-Jin;Kim, Han-Doo
    • Journal of Korea Multimedia Society
    • /
    • v.18 no.11
    • /
    • pp.1281-1288
    • /
    • 2015
  • Nongroup Cellular Automata(CA) having two trees in the state transition diagram of a CA is suitable for pattern classifier which divides pattern set into two classes. Maji et al. [1] classified patterns by using multiple attractor cellular automata as a pattern classifier with dependency vector. In this paper we propose a method of generation of a pattern classifier using feature vector which is the extension of dependency vector. In addition, we propose methods for finding nonreachable states in the 0-tree of the state transition diagram of TPMACA corresponding to the given feature vector for the analysis of the state transition behavior of the generated pattern classifier.