Automated Threshold Learning for Feature Selection Optimization

Koren, Michal; Peretz, Or; Koren, Oded

doi:10.2139/ssrn.4350765

Cited by 4 publications

(3 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The algorithm uses the automated threshold feature selection (AutoTFS), an automated method for feature selection in supervised learning (M. Koren, Peretz, et al, 2023). The defined method uses B (

B \geq 1

) for the feature selection and importance techniques.…”

Section: Weighted Classification Methodsmentioning

confidence: 99%

“…For each dataset described in Section 3.1, the following experiment was performed with the use of KNN: For AutoTFS,

T = 0.2

, and the different thresholds were compared between 0 and 0.2 with steps of 0.01, and the use of

α = 4

, as in the original article (M. Koren, Peretz, et al, 2023). For example, if

T = 0.1

and

α = 2

, all selected features must be at least 10% important for the model and selected by at least two different techniques. For the accuracy comparison, the data was split into train and test sets and used a test size of 25%. The number of neighbours (

k

) for KNN was defined as the input to the proposed method.…”

Section: Empirical Studymentioning

confidence: 99%

“…First, the AutoTFS method (M. Koren, Peretz, et al, 2023) was applied, creating

F^{'}

, a subset of 15 of the most significant features. Next, the neighbourhood of

x_{c}

was created with the top five nearest observations, including one normal , two suspect and two pathologic classes.…”

Section: Empirical Studymentioning

confidence: 99%

See 2 more Smart Citations

Weighted distance classification method based on data intelligence

Koren,

Peretz

2023

Expert Systems

Self Cite

View full text Add to dashboard Cite

In today's world, data is essential for enhancing an organization's development and decision‐making processes. Implementing artificial intelligence is necessary to analyse data and make meaningful recommendations. Machine learning distance classification methods are used to classify observations in various algorithms, such as K‐nearest neighbours (KNN), learning vector quantization and support vector machines, and are commonly used in academia and industry. However, this procedure faces a significant challenge in finding optimal parameters (i.e., distance metrics and the desired number of neighbours) in multidimensional datasets. This study presents a novel variation of a general method for classifying new observations. The method defines a new measure called closeness, which represents the proximity between an observation and the distribution. The advantages of this method are the use of both parametric and non‐parametric distance metrics and the ability to classify observations in cases where the simple method does not provide a clear answer. This method was demonstrated using KNN over three datasets and was observed to succeed in providing correct classifications, while the simple KNN method did not. The results showed that the proposed method increased the accuracy score to 40.7% in two of the three cases and that the closeness values were well defined by the proximity between the new observation and the given distribution. In addition, the F1 score increased up to 47.97%. The innovative method introduced here may be examined and used in various distance classification algorithms.

show abstract

B \geq 1

) for the feature selection and importance techniques.…”

Section: Weighted Classification Methodsmentioning

confidence: 99%

“…For each dataset described in Section 3.1, the following experiment was performed with the use of KNN: For AutoTFS,

T = 0.2

, and the different thresholds were compared between 0 and 0.2 with steps of 0.01, and the use of

α = 4

, as in the original article (M. Koren, Peretz, et al, 2023). For example, if

T = 0.1

and

α = 2

k

) for KNN was defined as the input to the proposed method.…”

Section: Empirical Studymentioning

confidence: 99%

See 1 more Smart Citation

Weighted distance classification method based on data intelligence

Koren,

Peretz

2023

Expert Systems

Self Cite

View full text Add to dashboard Cite

show abstract

An Automated and Goal-Oriented Clustering Procedure

Koren,

Peretz

2024

Lecture Notes in Networks and Systems

View full text Add to dashboard Cite

Clustering techniques are convenient tools for preparing and organizing unstructured and unclassified data. Depending on the data, they can be used to prepare for an analysis or to gain insight. However, choosing a clustering technique can be challenging when dealing with high-dimensional datasets. Most often, application requirements and data distribution need to be considered. Since clustering is defined as a complex problem to calculate, different algorithms may produce different results that meet the application's needs. This study presents an automated threshold-based and goal-oriented clustering procedure. It is based on the AutoML mechanism to estimate the most suitable hyperparameters according to predefined needs and can learn four clustering performance metrics thresholds for a given dataset. The significant advantages of this method are the automatic selection of clustering technique (i.e., partitional, hierarchical, density-based, or graph-based) and the ability to determine the output dynamically, according to predefined goals. We tested our method over four datasets and analyzed the results according to different goals. The results show that our method improved the silhouette score by 549.5% (from 0.105 to 0.682) compared to popular and commonly used K-means. Furthermore, clustering based on multiple metrics yielded more information than clustering by a single metric.

show abstract