• Journal of Computing Science and Engineering
• /
• 제10권4호
• /
• pp.118-127
• /
• 2016

An improved sample balanced genetic algorithm and Extreme Learning Machine (iSBGA-ELM) was designed for accurate diagnosis of Alzheimer disease (AD) and identification of biomarkers associated with AD in this paper. The proposed AD diagnosis approach uses a set of magnetic resonance imaging scans in Open Access Series of Imaging Studies (OASIS) public database to build an efficient AD classifier. The approach contains two steps: "voxels selection" based on an iSBGA and "AD classification" based on the ELM. In the first step, the proposed iSBGA searches for a robust subset of voxels with promising properties for further AD diagnosis. The robust subset of voxels chosen by iSBGA is then used to build an AD classifier based on the ELM. A robust subset of voxels keeps a high generalization performance of AD classification in various scenarios and highlights the importance of the chosen voxels for AD research. The AD classifier with maximum classification accuracy is created using an optimal subset of robust voxels. It represents the final AD diagnosis approach. Experiments with the proposed iSBGA-ELM using OASIS data set showed an average testing accuracy of 87%. Experiments clearly indicated the proposed iSBGA-ELM was efficient for AD diagnosis. It showed improvements over existing techniques.

• 한국정보통신학회논문지
• /
• 제11권3호
• /
• pp.619-626
• /
• 2007

건설적 귀납법은 사례들이 갖고 있는 속성들에 적합한 연산자를 적용하여 이들 사례들을 좀 더 효율적으로 분류할 수 있는 새로운 속성들을 도출해 내는 기법이다. 본 논문에서는 주어진 사례의 집합으로부터 PROSPECTOR에서 사용한 규칙 형태의 분류 규칙을 습득하는 유전 알고리즘 기반 귀납적 학습 환경을 위한 건설적 귀납법을 제시한다. 속성 결합 연산자와 유도된 속성의 유용성을 평가하기 위한 방법을 중심으로 건설적 귀납법에 대해 자세히 설명하고 다양한 사례 집합을 이용하여 건설적 귀납법이 유전 알고리즘 기반 학습 환경에 미치는 영향을 평가하였다.

• 인터넷정보학회논문지
• /
• 제24권6호
• /
• pp.1-11
• /
• 2023

Federated learning (FL) is a ground breaking machine learning paradigm that allow smultiple participants to collaboratively train models in a cloud environment, all while maintaining the privacy of their raw data. This approach is in valuable in applications involving sensitive or geographically distributed data. However, one of the challenges in FL is dealing with heterogeneous and non-independent and identically distributed (non-IID) data across participants, which can result in suboptimal model performance compared to traditionalmachine learning methods. To tackle this, we introduce FedGCD, a novel FL algorithm that employs Graph Neural Network (GNN)-based community detection to enhance model convergence in federated settings. In our experiments, FedGCD consistently outperformed existing FL algorithms in various scenarios: for instance, in a non-IID environment, it achieved an accuracy of 0.9113, a precision of 0.8798,and an F1-Score of 0.8972. In a semi-IID setting, it demonstrated the highest accuracy at 0.9315 and an impressive F1-Score of 0.9312. We also introduce a new metric, nonIIDness, to quantitatively measure the degree of data heterogeneity. Our results indicate that FedGCD not only addresses the challenges of data heterogeneity and non-IIDness but also sets new benchmarks for FL algorithms. The community detection approach adopted in FedGCD has broader implications, suggesting that it could be adapted for other distributed machine learning scenarios, thereby improving model performance and convergence across a range of applications.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제18권8호
• /
• pp.2067-2081
• /
• 2024

Recently, image analysis research has been actively conducted due to the accumulation of big image data and the development of deep learning. Image analytics research has different characteristics from other data such as data size, real-time, image quality diversity, structural complexity, and security issues. In addition, a large amount of data is required to effectively analyze images with deep-learning models. However, in many fields, the data that can be collected is limited, so there is a need for meta learning based image analysis technology that can effectively train models with a small amount of data. This paper presents a comprehensive survey of meta-learning-based object-tracking techniques. This approach comprehensively explores object tracking methods and research that can achieve high performance in data-limited situations, including key challenges and future directions. It provides useful information for researchers in the field and can provide insights into future research directions.

• 공학교육연구
• /
• 제27권4호
• /
• pp.3-11
• /
• 2024

This study classified domestic and international systems by type, presenting their key features and examples, with the aim of outlining future directions for system development and research. AI-based learning assistance systems can be categorized into instructional-learning evaluation types and academic recommendation types, depending on their purpose. Instructional-learning evaluation types measure learners' levels through initial diagnostic assessments, provide customized learning, and offer adaptive feedback visualized based on learners' misconceptions identified through learning data. Academic recommendation types provide personalized academic pathways and a variety of information and functions to assist with overall school life, based on the big data held by schools. Based on these characteristics, future system development should clearly define the development purpose from the planning stage, considering data ethics and stability, and should not only approach from a technological perspective but also sufficiently reflect educational contexts.

• 한국시스템다이내믹스연구
• /
• 제9권1호
• /
• pp.57-71
• /
• 2008

This research paper describes a complex and vague settings in which organization makes a decision and explains a role of decision maker's learning process. The original paper, written by Cohen, March, Olsen in 1972, said that all members of organization depended on the technology taken through trials and errors, which is the 'learning' process literally. But they intended to exclude the learning process in their simulation model because their PORTRAN model couldn't replicate the learning concept. As a result, they couldn't explain how all agents of garbage can simulation model resolve the problem dynamically. To overcome this original paper's limitations, we try to rebuild a learning process simulation model using by system dynamics approach that can capture the linkage between organization leanings and agents-based decision-makings. Our learning simulation results reveal two points. First, decision maker's leanings process improves the efficiency of decision making in complex situation. Second, group learning shows a superior efficiency to an individual learning because group members share organizational memory and energy.

• 한국게임학회 논문지
• /
• 제10권4호
• /
• pp.91-101
• /
• 2010

본 연구의 목적은 게임을 활용한 학습이 교육현장에서 교수-학습을 지원하는 방법으로 활용되기 위해 교사들의 인식을 조사하여 분석하는데 있다. 이는 게임기반학습의 확산을 위한 중요한 지표로 활용할 수 있을 것이다. 본 연구의 결과를 종합해보면, 첫째, 교사들이 게임을 사용하는데 있어 교육과정과 관련된 학습주제와 내용을 지닌 게임을 찾는데 어려움이 있다는 것이다. 둘째, 학습자의 다양한 수준에 따른 게임 활용 수업을 준비하는데 많은 시간이 소요된다는 것이다. 셋째, 교사가 게임을 수업에 활용하기 위해 관련기관에서 다양한 정보와 지침을 제공해야할 필요가 있다는 것이다. 마지막으로, 행-재정적인 지원이 체계적으로 이루어져야할 필요가 있는 것으로 나타났다.

• 제어로봇시스템학회논문지
• /
• 제13권4호
• /
• pp.358-364
• /
• 2007

Intelligent walking modeling of humanoid robot using learning based neuro-fuzzy system is presented in this paper. Walking pattern, trajectory of the zero moment point (ZMP) in a humanoid robot is used as an important criterion for the balance of the walking robots but its complex dynamics makes robot control difficult. In addition, it is difficult to generate stable and natural walking motion for a robot. To handle these difficulties and explain empirical laws of the humanoid robot, we are modeling practical humanoid robot using neuro-fuzzy system based on the two types of natural motions which are walking trajectories on a t1at floor and on an ascent. Learning based neuro-fuzzy system employed has good learning capability and computational performance. The results from neuro-fuzzy system are compared with previous approach.

• IEIE Transactions on Smart Processing and Computing
• /
• 제6권3호
• /
• pp.183-192
• /
• 2017

k-nearest neighbor (K-NN) is a well-known classification algorithm, being feature space-based on nearest-neighbor training examples in machine learning. However, K-NN, as we know, is a lazy learning method. Therefore, if a K-NN-based system very much depends on a huge amount of history data to achieve an accurate prediction result for a particular task, it gradually faces a processing-time performance-degradation problem. We have noticed that many researchers usually contemplate only classification accuracy. But estimation speed also plays an essential role in real-time prediction systems. To compensate for this weakness, this paper proposes correlation coefficient-based clustering (CCC) aimed at upgrading the performance of K-NN by leveraging processing-time speed and plurality rule-based density (PRD) to improve estimation accuracy. For experiments, we used real datasets (on breast cancer, breast tissue, heart, and the iris) from the University of California, Irvine (UCI) machine learning repository. Moreover, real traffic data collected from Ojana Junction, Route 58, Okinawa, Japan, was also utilized to lay bare the efficiency of this method. By using these datasets, we proved better processing-time performance with the new approach by comparing it with classical K-NN. Besides, via experiments on real-world datasets, we compared the prediction accuracy of our approach with density peaks clustering based on K-NN and principal component analysis (DPC-KNN-PCA).

• 한국지능시스템학회논문지
• /
• 제19권4호
• /
• pp.519-524
• /
• 2009

최근에 강화학습 기법은 기계학습 분야에서 많은 관심을 끌어왔다. 강화학습 관련 연구에서 가장 유력하게 사용되어 온 방법들로는 가치함수를 활용하는 기법, 제어규칙(policy) 탐색 기법 및 액터-크리틱 기법 등이 있는데, 본 논문에서는 이들 중 연속 상태 및 연속 입력을 갖는 문제를 위하여 액터-크리틱 기법의 틀에서 제안된 알고리즘들과 관련된 내용을 다룬다. 특히 본 논문은 퍼지 이론에 기반을 둔 액터-크리틱 계열 강화학습 기법인 ACFRL 알고리즘과, RLS 필터와 NAC(natural actor-critic) 기법에 기반을 둔 RLS-NAC 기법을 접목하는 방안을 집중적으로 고찰한다. 고찰된 방법론은 기는 로봇의 제어문제에 적용되고, 학습 성능의 비교로부터 얻어진 몇 가지 결과가 보고된다.