A kernel semi-supervised distance metric learning with relative distance: Integration with a MOO approach

Sanodiya, Rakesh Kumar; Saha, Sriparna; Mathew, Jimson

doi:10.1016/j.eswa.2018.12.051

Cited by 13 publications

(8 citation statements)

References 26 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The genetic algorithm developed is a modified version of the elitist non-dominated sorting genetic algorithm (NSGA) II (Deb et al (2002). NSGA II is efficient and used widely by many researchers in various disciplines (Alirezaei et al 2019;Avci and Selim 2017;Lahsasna and Seng 2017;Sanodiya et al 2019;Soui et al 2019;Wang et al 2018). It maintains diversity by seeking an even distribution of the non-dominated solutions in the objective space using the crowding distance (Deb et al 2002), a measure of the spatial density of the solutions in the objective space, based on the average distance between a solution and its nearest neighbours.…”

Section: Penalty-free Multi-objective Genetic Algorithmmentioning

confidence: 99%

Entropy maximizing evolutionary design optimization of water distribution networks under multiple operating conditions

Tanyimboh

Czajkowska

2021

Environ Syst Decis

View full text Add to dashboard Cite

The informational entropy model for flow networks was formulated over 30 years ago by Tanyimboh and Templeman (University of Liverpool, UK) for a single discrete operating condition that typically comprises the maximum daily demands and was undefined for water distribution networks (WDNs) under multiple operating conditions. Its extension to include multiple discrete operating conditions was investigated experimentally herein considering the relationships between flow entropy and hydraulic capacity reliability and redundancy. A novel penalty-free multi-objective genetic algorithm was developed to minimize the initial construction cost and maximize the flow entropy subject to the design constraints.Furthermore, optimized designs derived from the maximum daily demands as a single discrete operating condition were compared to those derived from a combination of discrete operating conditions. Optimized designs from a combination of discrete operating conditions outperformed those from a single operating condition in terms of performance and initial construction cost. The best results overall were achieved by maximizing the sum of the flow entropies of the discrete operating conditions. The logical inference from the results is that the flow entropy of multiple discrete operating conditions is the sum of their respective entropies. Also, a crucial property of the resulting flow entropy model is that it is bias free with respect to the individual operating conditions; hitherto a fundamental weakness concerning the practical application of the flow entropy model to WDNs is thus addressed.

show abstract

Section: Penalty-free Multi-objective Genetic Algorithmmentioning

confidence: 99%

Entropy maximizing evolutionary design optimization of water distribution networks under multiple operating conditions

Tanyimboh

Czajkowska

2021

Environ Syst Decis

View full text Add to dashboard Cite

show abstract

“…Machine learning algorithms can be used to further enhance a knowledge-intensive system [18]. This is because machine learning is applied in a multitude of tasks [19][20][21][22][23].…”

Section: Introductionmentioning

confidence: 99%

Implementation of Decision Tree Algorithm to Classify Knowledge Quality in a Knowledge Intensive System

et al. 2021

View full text Add to dashboard Cite

Knowledge is an important asset for an organisation as it facilitates organisational growth. To facilitate knowledge creation and sharing, this is where a knowledge-intensive system is required. One key area that hinders the effective use of knowledge-intensive systems in an organisation is the lack of knowledge quality. This causes the system to be underutilised, and as a result, knowledge will not be captured or shared effectively. Recent KM findings identified that machine learning could be beneficial to knowledge management. A literature review was conducted to identify knowledge of quality attributes and machine learning algorithms. From the findings, it was identified that the decision tree algorithm has a strong potential at classifying knowledge quality. An experiment was then devised to identify the training model required and measure its effectiveness using a pilot test. This involved using a knowledge-intensive system and mapping its variables to the respective knowledge quality attributes. From the experimentation result, the training model is then devised before implemented in a pilot test. The pilot test involved collecting knowledge using the same knowledge-intensive system before running the training model. From the results, it was identified that the decision tree could classify knowledge quality though the results yielded four different outputs at classifying knowledge quality. It was concluded that machine learning is beneficial in the area of knowledge management.

show abstract

“…The distances between data points in ML constraints should be minimum, while the distances between those in CL constraints must be large. But, according to [25,26], the relative-distance constraints like inequality constraints set (C neq ) and equality constraintsset (C eq ) are particularly effective in expressing structures at a finer level of detail when compared to ML and CL constraints. Therefore, in our proposed work, we consider relative distance constraints as side information additionally to the feature vector for preserving the discriminate information.…”

Section: Introductionmentioning

confidence: 99%

Unsupervised Transfer Learning via Relative Distance Comparisons

Sanodiya

Yao

2020

IEEE Access

Self Cite

View full text Add to dashboard Cite

Primitive machine learning method such as Support Vector Machine (SVM) or k-Nearest Neighbor (k-NN) faces a major challenge when its training and test data is distributed with large-scale variations in lighting conditions, color, backgrounds, size, etc. The variation may be because training and testing data can come from related but some other domains. Considerable efforts have been made in the development of transfer learning methods. However, most current work focuses only on the following goals or objectives: i)Preservation of source domain discriminative information with Linear Discriminant Analysis (LDA); ii)Maximization of target domain variance; iii) Subspace alignment; iv) Minimization of marginal and conditional distribution by using the Maximum Mean Discrepancy (MMD) criterion;v)Preservation of original similarity of the data samples. Current approaches to preserve source domain discriminant information can easily misclassify the target domain samples which are distributed near the edge of the cluster. In order to overcome the limitations of existing transfer learning methods, we propose a novel Unsupervised Transfer Learning Via Relative Distance Comparisons (UTRDC) method. UTRDC optimizes all the aforementioned objectives jointly with a common projection vector matrix for both domains as well as uses the relative distance constraints for better inter-class separability and intra-class compactness. Furthermore, we extend our proposed method UTRDC to a kernelized version to deal with non-linear separable datasets. Extensive experimentation on two real-world problems datasets (PIE face and Office+Caltech) has proven that the proposed methods outperform several approaches to non-transfer learning and transfer learning.

show abstract

A kernel semi-supervised distance metric learning with relative distance: Integration with a MOO approach

Cited by 13 publications

References 26 publications

Entropy maximizing evolutionary design optimization of water distribution networks under multiple operating conditions

Entropy maximizing evolutionary design optimization of water distribution networks under multiple operating conditions

Implementation of Decision Tree Algorithm to Classify Knowledge Quality in a Knowledge Intensive System

Unsupervised Transfer Learning via Relative Distance Comparisons

Contact Info

Product

Resources

About