Supervised outlier detection for classification and regression

Fernández, Ángela; Bella, Juan; Dorronsoro, José R.

doi:10.1016/j.neucom.2022.02.047

Cited by 30 publications

(10 citation statements)

References 28 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Outlier detection (OD) is an important step in maximizing performance of an ML algorithm [ 50 ]. To remove outliers in the dataset, we performed the random sampling and consensus (RANSAC) method as has been applied in regression problems [ 51 ].…”

Section: Methodsmentioning

confidence: 99%

Classifying Malignancy in Prostate Glandular Structures from Biopsy Scans with Deep Learning

Fogarty

Goldgof

Hall

et al. 2023

Cancers

View full text Add to dashboard Cite

Histopathological classification in prostate cancer remains a challenge with high dependence on the expert practitioner. We develop a deep learning (DL) model to identify the most prominent Gleason pattern in a highly curated data cohort and validate it on an independent dataset. The histology images are partitioned in tiles (14,509) and are curated by an expert to identify individual glandular structures with assigned primary Gleason pattern grades. We use transfer learning and fine-tuning approaches to compare several deep neural network architectures that are trained on a corpus of camera images (ImageNet) and tuned with histology examples to be context appropriate for histopathological discrimination with small samples. In our study, the best DL network is able to discriminate cancer grade (GS3/4) from benign with an accuracy of 91%, F1-score of 0.91 and AUC 0.96 in a baseline test (52 patients), while the cancer grade discrimination of the GS3 from GS4 had an accuracy of 68% and AUC of 0.71 (40 patients).

show abstract

Section: Methodsmentioning

confidence: 99%

Classifying Malignancy in Prostate Glandular Structures from Biopsy Scans with Deep Learning

Fogarty

Goldgof

Hall

et al. 2023

Cancers

View full text Add to dashboard Cite

show abstract

“…Third, this study used supervised machine learning models for binary classification of students at risk in mathematics and those who are not. Future studies can utilize supervised anomaly detection using machine learning and deep learning algorithms (see Ángela Fernández et al [45] for a review of different algorithms) to identify abnormal patterns (e.g., students who perform significantly worse than their peers). Fourth, depending on the outcomes of the screening procedure, students could be more or less frequently tested based on their academic growth in mathematics throughout the academic year.…”

Section: Limitations and Future Directionsmentioning

confidence: 99%

Optimized Screening for At-Risk Students in Mathematics: A Machine Learning Approach

2022

View full text Add to dashboard Cite

Traditional screening approaches identify students who might be at risk for academic problems based on how they perform on a single screening measure. However, using multiple screening measures may improve accuracy when identifying at-risk students. The advent of machine learning algorithms has allowed researchers to consider using advanced predictive models to identify at-risk students. The purpose of this study is to investigate if machine learning algorithms can strengthen the accuracy of predictions made from progress monitoring data to classify students as at risk for low mathematics performance. This study used a sample of first-grade students who completed a series of computerized formative assessments (Star Math, Star Reading, and Star Early Literacy) during the 2016–2017 (n = 45,478) and 2017–2018 (n = 45,501) school years. Predictive models using two machine learning algorithms (i.e., Random Forest and LogitBoost) were constructed to identify students at risk for low mathematics performance. The classification results were evaluated using evaluation metrics of accuracy, sensitivity, specificity, F1, and Matthews correlation coefficient. Across the five metrics, a multi-measure screening procedure involving mathematics, reading, and early literacy scores generally outperformed single-measure approaches relying solely on mathematics scores. These findings suggest that educators may be able to use a cluster of measures administered once at the beginning of the school year to screen their first grade for at-risk math performance.

show abstract

“…An improved content-based outlier detection method was proposed in [11]. In [12], a novel supervised outlier estimator was designed. This was performed by pipelining an outlier detector in such a manner that the targets involved in the outlier detector were obtained in an optimal manner.…”

Section: Related Workmentioning

confidence: 99%

Gaussian Projection Deep Extreme Clustering and Chebyshev Reflective Correlation based Outlier Detection

Rajalakshmi¹,

Madhubala²

2022

IJACSA

View full text Add to dashboard Cite

Outlier detection or simply the task of point detection that are noticeably distinct and different from data sample is a predominant issue in deep learning. When a framework is constructed, these distinctive points can later lead to model training and compromise accurate predictions. Owing to this reason, it is paramount to recognize and eliminate them before constructing any supervised model and this is frequently the initial step when dealing with a deep learning issue. Over the recent few years, different numbers of outlier detector algorithms have been designed that ensure satisfactory results. However, their main disadvantages remain in the time and space complexity and unsupervised nature. In this work, a clusteringbased outlier detection called, Random Projection Deep Extreme Learning-based Chebyshev Reflective Correlation (RPDEL-CRC) is proposed. First, Gaussian Random Projection-based Deep Extreme Learning-based Clustering model is designed. Here, by applying Gaussian Random Projection function to the Deep Extreme Learning obtains the relevant and robust clusters corresponding to the data points in a significant manner. Next, with the robust clusters, outlier detection time is said to be reduced to a greater extent. In addition, a novel Chebyshev Temporal and Reflective Correlation-based Outlier Detection model is proposed to detect outliers therefore achieving high outlier detection accuracy. The proposed approach is validated with the NIFTY-50 stock market dataset. The performance of the RPDEL-CRC method is evaluated by applying it to NIFTY-50 Stock Market dataset. Finally, we compare the results of the RPDEL-CRC method to the state-of-the-art outlier detection methods using outlier detection time, accuracy, error rate and false positive rate evaluation metrics.

show abstract

Supervised outlier detection for classification and regression

Cited by 30 publications

References 28 publications

Classifying Malignancy in Prostate Glandular Structures from Biopsy Scans with Deep Learning

Classifying Malignancy in Prostate Glandular Structures from Biopsy Scans with Deep Learning

Optimized Screening for At-Risk Students in Mathematics: A Machine Learning Approach

Gaussian Projection Deep Extreme Clustering and Chebyshev Reflective Correlation based Outlier Detection

Contact Info

Product

Resources

About