New multivariate kernel density estimator for uncertain data classification

Kim, Byung–Hoon; Jeong, Young Seon; Jeong, Myong K.

doi:10.1007/s10479-020-03715-4

Cited by 5 publications

(4 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To convert the numerical data into probabilistic values, a KDE-based transformation was implemented. The use of KDE for feature transformation has shown better performance than without KDE-based transformation [ 28 ]. Here, for each feature, a KDE was generated using the group of patients with polyps, then the KDE was normalized to 1, and an exponent between 1 and 4 was applied (Fig.…”

Section: Resultsmentioning

confidence: 99%

Artificial intelligence for the prevention and prediction of colorectal neoplasms

Tokutake

Morelos-Gómez²,

Hoshi

et al. 2023

J Transl Med

View full text Add to dashboard Cite

Background Colonoscopy is a useful as a cancer screening test. However, in countries with limited medical resources, there are restrictions on the widespread use of endoscopy. Non-invasive screening methods to determine whether a patient requires a colonoscopy are thus desired. Here, we investigated whether artificial intelligence (AI) can predict colorectal neoplasia. Methods We used data from physical exams and blood analyses to determine the incidence of colorectal polyp. However, these features exhibit highly overlapping classes. The use of a kernel density estimator (KDE)-based transformation improved the separability of both classes. Results Along with an adequate polyp size threshold, the optimal machine learning (ML) models’ performance provided 0.37 and 0.39 Matthews correlation coefficient (MCC) for the datasets of men and women, respectively. The models exhibit a higher discrimination than fecal occult blood test with 0.047 and 0.074 MCC for men and women, respectively. Conclusion The ML model can be chosen according to the desired polyp size discrimination threshold, may suggest further colorectal screening, and possible adenoma size. The KDE feature transformation could serve to score each biomarker and background factors (health lifestyles) to suggest measures to be taken against colorectal adenoma growth. All the information that the AI model provides can lower the workload for healthcare providers and be implemented in health care systems with scarce resources. Furthermore, risk stratification may help us to optimize the efficiency of resources for screening colonoscopy.

show abstract

Section: Resultsmentioning

confidence: 99%

Artificial intelligence for the prevention and prediction of colorectal neoplasms

Tokutake

Morelos-Gómez²,

Hoshi

et al. 2023

J Transl Med

View full text Add to dashboard Cite

show abstract

“…As for practical significance, the proposed CVIs could be utilized in diverse applications. For example, Kim et al proposed new a multivariate kernel density estimator for uncertain data classification for mixed defect patterns on DRAM wafer maps [ 31 ]. The proposed CVI method could be applied for evaluating the number of defect patterns on wafer maps.…”

Section: Discussionmentioning

confidence: 99%

Cluster Validity Index for Uncertain Data Based on a Probabilistic Distance Measure in Feature Space

Baek

Tavakkol

et al. 2023

Sensors

Self Cite

View full text Add to dashboard Cite

Cluster validity indices (CVIs) for evaluating the result of the optimal number of clusters are critical measures in clustering problems. Most CVIs are designed for typical data-type objects called certain data objects. Certain data objects only have a singular value and include no uncertainty, so they are assumed to be information-abundant in the real world. In this study, new CVIs for uncertain data, based on kernel probabilistic distance measures to calculate the distance between two distributions in feature space, are proposed for uncertain clusters with arbitrary shapes, sub-clusters, and noise in objects. By transforming original uncertain data into kernel spaces, the proposed CVI accurately measures the compactness and separability of a cluster for arbitrary cluster shapes and is robust to noise and outliers in a cluster. The proposed CVI was evaluated for diverse types of simulated and real-life uncertain objects, confirming that the proposed validity indexes in feature space outperform the pre-existing ones in the original space.

show abstract

“…Physiological systems are known to have a high degree of dependence with regard to a stress response, because they are often initiated by the same neuroendocrine axis [10]. Some researchers have shown classifiers may account for dependencies using multivariate kernel density estimators [11]. Therefore, it may be beneficial to evaluate supervised machine learning classifiers against a benchmark optimal classifier that approximates Bayes using a density distribution estimated through multi-variate kernel density estimation for stress detection.…”

Section: Introductionmentioning

confidence: 99%

“…Real-time stress detection could enable closed-loop automation to either modify the training environments to better match the trainee's responses or better assess individual stress during staged or real operations [13]. In datasets with repeated measurements at multiple times that present uncertainty from randomness or incompleteness, such as multiple measures of physiological data, multivariate kernel density estimators may help increase detection accuracy [11].…”

Section: Introductionmentioning

confidence: 99%

Real-Time Personalized Physiologically Based Stress Detection for Hazardous Operations

et al. 2023

View full text Add to dashboard Cite

When training for hazardous operations, real-time stress detection is an asset for optimizing task performance and reducing stress. Stress detection systems train a machine-learning model with physiological signals to classify stress levels of unseen data. Unfortunately, individual differences and the time-series nature of physiological signals limit the effectiveness of generalized models and hinder both posthoc stress detection and real-time monitoring. This study evaluated a personalized stress detection system that selects a personalized subset of features for model training. The system was evaluated post-hoc for realtime deployment. Further, traditional classifiers were assessed for error caused by indirect approximations against a benchmark, optimal probability classifier (Approximate Bayes; ABayes). Healthy participants completed a task with three levels of stressors (low, medium, high), either a complex task in virtual reality (responding to spaceflight emergency fires, n=27) or a simple laboratory-based task (N-back, n=14). Heart rate, blood pressure, electrodermal activity, and respiration were assessed. Personalized features and different interval window sizes were compared. Classification performance was compared for ABayes, support vector machine, decision tree, and random forest. The results demonstrate that a personalized model with time series intervals can classify three stress levels with high accuracy better than a generalized model. However, crossvalidation and holdout performance varied for traditional classifiers vs. ABayes, suggesting error from indirect approximations. The selected features changed with windows size and differed between tasks, but blood pressure contributed to the most prominent features. The capability to account for individual difference is a prominent advantage of personalized models, and will likely have a growing presence in future detection systems.

show abstract

New multivariate kernel density estimator for uncertain data classification

Cited by 5 publications

References 22 publications

Artificial intelligence for the prevention and prediction of colorectal neoplasms

Artificial intelligence for the prevention and prediction of colorectal neoplasms

Cluster Validity Index for Uncertain Data Based on a Probabilistic Distance Measure in Feature Space

Real-Time Personalized Physiologically Based Stress Detection for Hazardous Operations

Contact Info

Product

Resources

About