Recent advances in scaling‐down sampling methods in machine learning

ElRafey, Amr; Wojtusiak, Janusz

doi:10.1002/wics.1414

Cited by 24 publications

(13 citation statements)

References 122 publications

(130 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As it has been observed in the review of [18], data sampling methods for machine learning have been investigated for decades. According to the above paper, in recent years progress has been made in methods that can be broadly categorized into random sampling including density-biased and nonuniform sampling methods, active learning methods, which are the type of semisupervised learning, and progressive sampling methods, which can be viewed as a combination of the above two approaches.…”

Section: Related Workmentioning

confidence: 99%

Incremental Gene Expression Programming Classifier with Metagenes and Data Reduction

Jȩdrzejowicz

Jędrzejowicz

2018

Complexity

View full text Add to dashboard Cite

The paper proposes an incremental Gene Expression Programming classifier. Its main features include using two-level ensemble consisting of base classifiers in form of genes and the upper-level classifier in the form of metagene. The approach enables us to deal with big datasets through controlling computation time using data reduction mechanisms. The user can control the number of attributes used to induce base classifiers as well as the number of base classifiers used to induce metagenes. To optimize the parameter setting phase, an approach based on the Orthogonal Experiment Design principles is proposed, allowing for statistical evaluation of the influence of different factors on the classifier performance. In addition, the algorithm is equipped with a simple mechanism for drift detection. A detailed description of the algorithm is followed by the extensive computational experiment. Its results validate the approach. Computational experiment results show that the proposed approach compares favourably with several state-of-the-art incremental classifiers.

show abstract

Section: Related Workmentioning

confidence: 99%

Incremental Gene Expression Programming Classifier with Metagenes and Data Reduction

Jȩdrzejowicz

Jędrzejowicz

2018

Complexity

View full text Add to dashboard Cite

show abstract

“…In the case of optimization, the experiment can be conducted using a selected optimization algorithm, which has been a mature field. On the other hand, in the case of constructing an accurate model for advanced process development or predictive maintenance, data-efficient sampling in semiconductor manufacturing has been less studied compared to optimization, while sampling has been an intensively studied subject in other fields. − Prior works in data-efficient sampling in the semiconductor process are mainly in yield improvement, quality control, and predictive maintenance in a production-line setting. − On the other hand, the sampling strategies in developing key process steps in advanced technology nodes have not been studied much, and we only found limited literature. , …”

Section: Introductionmentioning

confidence: 99%

“…Thus, an active learning algorithm manages to gain examples that can immediately minimize the number of hypotheses, and the two most widely used methods are query by disagreement and query by committee. , In the variance reduction method, the active data selection is used to minimize the predictive variance of the active learning algorithm depending on Fisher information. Nevertheless, it will face a detriment that the computational complexity makes it impractical since a large number of parameters are present. ,− Hence, it has received little interest in actual applications of efficient sampling in ML problems.…”

Section: Introductionmentioning

confidence: 99%

“…Passive sampling chooses samples based on the geometry of the features, while active sampling selects the next samples based on the output of the model trained by the currently available data . Three methods are used in active learning: uncertainty sampling, minimizing the hypothesis space, and variance reduction. , In uncertainty sampling, several methods exist to conduct uncertainty samplings, such as least confidence and margin confidence. The least confidence method chooses data whose confidence is less than a specific number defined by the user.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Meta-Learned and TCAD-Assisted Sampling in Semiconductor Laser Annealing

et al. 2022

View full text Add to dashboard Cite

While applying machine learning (ML) to semiconductor manufacturing is prevalent, an efficient way to sample the search space has not been explored much in key processes such as lithography, annealing, deposition, and etching. The aim is to use the fewest experimental trials to construct an accurate predictive model. Here, we proposed a technology computer added design (TCAD)-assisted meta-learned sampling approach. The meta-learner adjusts the way of sampling in terms of how to hybridize the TCAD with ML when selecting the next sampling point. While an advanced semiconductor process is expensive, efficient sampling is indispensable. Using laser annealing as an example, we demonstrate the effectiveness of the proposed algorithm where the mean square error (MSE) at the first 100 sampling steps using TCAD-assisted meta-learned sampling is significantly lower than the pure ML approach. Besides, with reference to the pure TCAD approach, the TCAD-assisted sampling prevents the MSE degradation at 200−400 sampling steps. The proposed approach can be used in other manufacturing or even any applied machine intelligence fields.

show abstract

“…Variance Reduction Technique (VRT) is proposed as the most effective approach to improve sampling efficiency [13,14]. Although VRT increases sampling efficiency, it does not use the minimum sample size for UA [15]. Progressive Sampling Techniques (PST) solves the problem of finding the least possible samples.…”

Section: Introductionmentioning

confidence: 99%

Developing a Multi-objective Multi-Disciplinary Robust Design Optimization Framework

2021

View full text Add to dashboard Cite

The purpose of this study is to provide an efficient Multi-Objective Multidisciplinary Robust Design Optimization (MOMRDO) framework. To this end, Bi-Level Integrated System Synthesis (BLISS) framework is implemented as a fast Multi-disciplinary Design Optimization (MDO) framework. Progressive Latin Hypercube Sampling (PLHS) is developed as a Design of Experiment (DOE) of the Uncertainty Analysis (UA). This systematic approach leads to a fast, adaptive and efficient framework for Robust Design Optimization (RDO) of complex systems. The accuracy and performance of the proposed algorithm have been evaluated with various tests. Finally, the RDO of a hydrazine monopropellant thruster is defined as a case study. The results show that the proposed method is a fast and efficient method for the multi-objective optimization design of complex systems, and this approach can be used for other engineering applications as well.

show abstract

Recent advances in scaling‐down sampling methods in machine learning

Cited by 24 publications

References 122 publications

Incremental Gene Expression Programming Classifier with Metagenes and Data Reduction

Incremental Gene Expression Programming Classifier with Metagenes and Data Reduction

Meta-Learned and TCAD-Assisted Sampling in Semiconductor Laser Annealing

Developing a Multi-objective Multi-Disciplinary Robust Design Optimization Framework

Contact Info

Product

Resources

About