RETRACTED ARTICLE: A taxonomy on impact of label noise and feature noise using machine learning techniques

Shanthini, A.; Vinodhini, G.; Chandrasekaran, R. M.; Supraja, P.

doi:10.1007/s00500-019-03968-7

Cited by 24 publications

(19 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…By operating in an agnostic manner to capture non-linear multi-dimensional interactions and infer the degree of class ownership, these tools may be better suited to explain the distinctions between structural classes than hard, binary decision-boundaries set by a-priori assumptions in classical hypothesis-testing approaches (Rutledge et al 2019 ; Li and Tong 2020 ). Decision-tree ML models are particularly favourable when seeking to explain variables of interests from non-normally distributed data such as self-reported independent subjective experiences and derive good explanatory value even in the presence of major scoring noise (Shanthini et al 2019 ). Together with their redeployable nature once trained, they be may useful tools to generalise measures of subjective effects.…”

Section: Introductionmentioning

confidence: 99%

The use patterns of novel psychedelics: experiential fingerprints of substituted phenethylamines, tryptamines and lysergamides

et al. 2022

View full text Add to dashboard Cite

Background Novel psychedelics (NPs) are an expanding set of compounds, presenting new challenges for drug policy and opportunities for clinical research. Unlike their classical derivatives, little is known regarding their use profiles or their subjective effects. Aims The purpose of this study was to compile usage patterns and adverse event rates for individual NPs belonging to each of three main psychedelic structural families. Targeting the most widely used representatives for each class, we expanded on their phenomenological distinctions. Methods A two-part survey was employed. We investigated the prevalence of novel phenethylamines, tryptamine and lysergamides in NP users (N = 1180), contrasting the type and incidence of adverse events (AEs) using a set of logistic regressions. Honing in on 2–4-Bromo-2,5-dimethoxyphenyl)ethanamine (2C-B) (48.6%), 1-propionyl-lysergic acid diethylamide (1P-LSD) (34.2%) and 4-Acetoxy-N,N-dimethyltryptamine (4-AcO-DMT) (23.1%), we examined their phenomenological separability using a gradient boosting (XGBoost) supervised classifier. Results Novel phenethylamines had the highest prevalence of use (61.5%) seconded by tryptamines (43.8%) and lysergamides (42.9%). Usage patterns were identified for 32 different compounds, demonstrating variable dosages, durations and a common oral route of administration. Compared to phenethylamines, the odds for tryptamines and lysergamides users were significantly less for overall physical AEs. No significant differences in overall psychological AEs were found. Overall model area under the curve (AUC) stood at 0.79 with sensitivity (50.0%) and specificity (60.0%) for 2C-B ranking lowest. Conclusion NP classes may hold distinct AE rates and phenomenology, the latter potentially clouded by the subjective nature of these experiences. Further targeted research is warranted.

show abstract

Section: Introductionmentioning

confidence: 99%

The use patterns of novel psychedelics: experiential fingerprints of substituted phenethylamines, tryptamines and lysergamides

et al. 2022

View full text Add to dashboard Cite

show abstract

“…The attack impact of the entropy approach becomes more visible as the noise level rises, and it outperforms other methods, according to the data. In Shanthini et al's study, 51 the robustness of three boosting learners was tested on three medical data sets under the effect of features and label noise. The findings of the experiments reveal that label noise does significantly more harm than feature noise.…”

Section: Label-flipping Poisoning Attackmentioning

confidence: 99%

The robustness of popular multiclass machine learning models against poisoning attacks: Lessons and insights

Maabreh

Qolomany

et al. 2022

International Journal of Distributed Sensor Networks

View full text Add to dashboard Cite

Despite the encouraging outcomes of machine learning and artificial intelligence applications, the safety of artificial intelligence–based systems is one of the most severe challenges that need further exploration. Data set poisoning is a severe problem that may lead to the corruption of machine learning models. The attacker injects data into the data set that are faulty or mislabeled by flipping the actual labels into the incorrect ones. The word “robustness” refers to a machine learning algorithm’s ability to cope with hostile situations. Here, instead of flipping the labels randomly, we use the clustering approach to choose the training samples for label changes to influence the classifiers’ performance and the distance-based anomaly detection capacity in quarantining the poisoned samples. According to our experiments on a benchmark data set, random label flipping may have a short-term negative impact on the classifier’s accuracy. Yet, an anomaly filter would discover on average 63% of them. On the contrary, the proposed clustering-based flipping might inject dormant poisoned samples until the number of poisoned samples is enough to influence the classifiers’ performance severely; on average, the same anomaly filter would discover 25% of them. We also highlight important lessons and observations during this experiment about the performance and robustness of popular multiclass learners against training data set–poisoning attacks that include: trade-offs, complexity, categories, poisoning resistance, and hyperparameter optimization.

show abstract

“…[12] also raised the question on the quality of data of NASA repository, but still, a few powerful machine learning-based defect prediction models are available such as [19,43,44,45]. There are two different types of noises in a defect dataset, and both of these noises [46] affect the performance over machine learning algorithms, first is class noise and second is feature noise. However, we have only considered class noise in this article.…”

Section: Quality Of Defect Datasetmentioning

confidence: 99%

An empirical study toward dealing with noise and class imbalance issues in software defect prediction

Pandey

Tripathi

2021

Soft Comput

View full text Add to dashboard Cite

show abstract

RETRACTED ARTICLE: A taxonomy on impact of label noise and feature noise using machine learning techniques

Cited by 24 publications

References 19 publications

The use patterns of novel psychedelics: experiential fingerprints of substituted phenethylamines, tryptamines and lysergamides

The use patterns of novel psychedelics: experiential fingerprints of substituted phenethylamines, tryptamines and lysergamides

The robustness of popular multiclass machine learning models against poisoning attacks: Lessons and insights

An empirical study toward dealing with noise and class imbalance issues in software defect prediction

Contact Info

Product

Resources

About