STRIP - a strip-based neural-network growth algorithm for learning multiple-valued functions

Ngom, Alioune; Stojmenović, Ivan; Milutinović, Veljko

doi:10.1109/72.914519

Cited by 24 publications

(9 citation statements)

References 33 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…If algorithms need acceleration, due to Big Data, possibilities are: (a) The existing computing architecture could be enhanced [24], (b) A new computing architecture could be introduced [25], (c) The number of iterations of the algorithm could be minimized using machine intelligence [26], or (d) Each iteration could be cut shorter if some kind of suboptimal computing is utilized [27].…”

Section: The Fourth Casementioning

confidence: 99%

Conflict resolution in the case of convective weather cell circumvention

Bogdanović

Mao

2023

J Big Data

View full text Add to dashboard Cite

This research analyzes the area required for the conflict resolution between aircraft in two flows impacted by a convective weather cell (CWC). The CWC is introduced as a constrained area, forbidden to flight through, which affects the air traffic flows. Prior the conflict resolution, two flows and their intersection are relocated away from the CWC area (thus enabling circumvention of the CWC), which is followed by a tuning of the relocated flows intersection angle in order to create the minimal size of the conflict zone (CZ—a circular area centered at the intersection of two flows, which provides aircraft enough space to completely resolve the conflict within). Therefore, the essence of the proposed solution is in providing conflict free trajectories for the aircraft in intersecting flows that are affected by the CWC, with the goal of minimizing the CZ size, so the finite occupied airspace for the conflict resolution and the CWC circumvention could be reduced. Compared to the best solutions and current industry practice, this article is focused in reduction of the airspace required for aircraft to aircraft and aircraft to weather conflict resolution, and not to distance travelled, time savings, and fuel consumption minimization. The conducted analysis in the MicrosoftExcel2010 confirmed the relevance of the proposed model and demonstrated variations in efficiency of the utilized airspace. The proposed model's transdisciplinary nature makes it potentially applicable in other fields of study, such as the conflict resolution between unmanned aerial vehicles (UAVs) and fixed objects like buildings. Building on this model and taking in consideration large and complex data sets, such as weather related data and flight data (aircraft position, speed, and altitude), we believe it is possible to conduct more sophisticated analyses that would take advantage of Big Data.

show abstract

Section: The Fourth Casementioning

confidence: 99%

Conflict resolution in the case of convective weather cell circumvention

Bogdanović

Mao

2023

J Big Data

View full text Add to dashboard Cite

show abstract

“…Model selection depends on algorithm selection, hyperparameter optimization and data preprocessing, like reduction and integration, which are often heuristically found by the systematic grid, a random search, or by applying machine learning [42], and this paper addresses algorithm selection only. Using logical domain-specific meta-features would make it possible to efficiently provide results for a given task and budget by minimizing computational resources and amount of data required for model selection in AutoML frameworks, which would speed up the learning process and improve overall performance.…”

Section: B Problem Analysismentioning

confidence: 99%

Novel Meta-Features for Automated Machine Learning Model Selection in Anomaly Detection

et al. 2021

View full text Add to dashboard Cite

A growing number of research papers shed light on automated machine learning (AutoML) frameworks, which are becoming a promising solution for building complex machine learning models without human expertise and assistance. The key challenge in enabling AutoML frameworks to build an efficient model for anomaly detection tasks is to determine the best underlying model for a given task and optimization metric. The meta-learning approaches based on a set of meta-features that describes data properties can enable efficient model selection in AutoML frameworks. The existing meta-learning approaches based on statistical and information-theoretic meta-features require large amounts of data and computational resources to extract data properties. This paper proposes a novel set of meta-features for model selection in anomaly detection tasks based on domain-specific properties of data which overcomes the shortcomings of existing meta-features by introducing simple but effective meta-features that can be efficiently extracted or estimated by using a low amount of data. Experiments with 63 datasets from different repositories with varying schemas show that the proposed set of meta-features achieves an accuracy of 87% for model selection, while the achieved accuracy for simple meta-features is 74%, for statistical meta-features 68%, for information theory meta-feature 70%, and for a comprehensive set of meta-features by pyMFE 73%. This demonstrates that the proposed set can be adopted by AutoML frameworks across a diverse range of domains.

show abstract

“…There are also other works, see Maniezzo (1994), Mitra and Hayashi (2000) and Pattichis and Schizas (1996), for their attempt in providing an exhaustive survey of neuro‐fuzzy rule generation algorithms. Moreover, Ngom et al (2001) considered the problem of synthesizing multiple‐valued logic functions by NNs, and the authors described a GA which finds the longest strip in. Most recently, Palaniappan et al (2002) applied NNs to classify alcoholics and non‐alcoholics by features extracted from visual evoked potential (VEP).…”

Section: Related Literature Reviewmentioning

confidence: 99%

Shape reconstruction by genetic algorithms and artificial neural networks

Liu

Tang

Frazer

2003

Engineering Computations

View full text Add to dashboard Cite

This paper presents a new surface reconstruction method based on complex form functions, genetic algorithms and neural networks. Surfaces can be reconstructed in an analytical representation format. This representation is optimal in the sense of least‐square fitting by predefined subsets of data points. The surface representations are achieved by evolution via repetitive application of crossover and mutation operations together with a back‐propagation algorithm until a termination condition is met. The expression is finally classified into specific combinations of basic functions. The proposed method can be used for CAD model reconstruction of 3D objects and free smooth shape modelling. We have implemented the system demonstration with Visual C++ and MatLab to enable real time surface visualisation in the process of design.

show abstract

STRIP - a strip-based neural-network growth algorithm for learning multiple-valued functions

Cited by 24 publications

References 33 publications

Conflict resolution in the case of convective weather cell circumvention

Conflict resolution in the case of convective weather cell circumvention

Novel Meta-Features for Automated Machine Learning Model Selection in Anomaly Detection

Shape reconstruction by genetic algorithms and artificial neural networks

Contact Info

Product

Resources

About