What question did this study set out to answer?

The aim is to identify methods for detecting anomalous data and assess their impact on data quality in AI models.

March 1, 2026Optical Memory and Neural Networks1 citations

Methods of Anomalous Data Detection in Datasets

ASAkbar SolievSamarkand State University named after Sharof Rashidov AAAkmal AkhatovSamarkand State University named after Sharof Rashidov ARAkbar RashidovKent Hastanesi

Key Points

The aim is to identify methods for detecting anomalous data and assess their impact on data quality in AI models.
Review of various anomaly detection techniques
Analysis of the impact of anomalous data on AI performance
Exploration of use cases for different detection methods
Identified that anomalous data impacts AI model accuracy significantly
Outlined various types of anomalies and their implications
Demonstrated that existing models may misinterpret data with anomalies as valid.

Abstract

It is known that the accuracy of data analysis and artificial intelligence models that trained and tuned on the basis of data is closely related to the quality of the data set. The quality of the data set depends on several factors, one of the most important of which is the absence or elimination of anomalous data in the data set. Anomalous data has such a property that artificial intelligence models work normally with a data set with this anomalous data. That is, artificial intelligence models do not notice at all that they are working with incorrect data. As a result, the artificial intelligence model returns an incorrect result, which may lead to incorrect conclusions about the object. Therefore, today, the detection of anomalous data in the datasets is one of the studies that has retained its relevance. This research paper discusses anomalous data, their negative consequences, and the types of anomalies in the data set. It also studies methods for detecting anomalous data in datasets and analyzes their use cases.

Ask AI

Helpful

Bookmark

View Full Paper