Outlier detection by active learning

Abe, Naoki; Zadrozny, Bianca; Langford, John

doi:10.1145/1150402.1150459

Cited by 260 publications

(192 citation statements)

References 12 publications

Supporting

Mentioning

187

Contrasting

Unclassified

Order By: Relevance

“…It is basically used in fraud detection, marketing and scientific discovery. Data mining actually refers to extracting the hidden interesting patterns from the large amount of datasets and databases [2]. Mining is basically used to uncover the patterns of the data, but this can be carried out on the sample of data.…”

Section: Introductionmentioning

confidence: 99%

Comparative Analysis of Outlier Detection Techniques

Malik¹,

Sadawarti²,

S³

2014

IJCA

View full text Add to dashboard Cite

Data Mining simply refers to the extraction of very interesting patterns of the data from the massive data sets. Outlier detection is one of the important aspects of data mining which actually finds out the observations that are deviating from the common expected behavior. Outlier detection and analysis is sometimes known as outlier mining. In this paper, we have tried to provide the broad and a comprehensive literature survey of outliers and outlier detection techniques under one roof, so as to explain the richness and complexity associated with each outlier detection technique. Moreover, we have also given a broad comparison of the various methods of the different outlier techniques.

show abstract

Section: Introductionmentioning

confidence: 99%

Comparative Analysis of Outlier Detection Techniques

Malik¹,

Sadawarti²,

S³

2014

IJCA

View full text Add to dashboard Cite

show abstract

“…Another issue is to find the optimized labels for the anomaly class which is really challenging. It was addressed in [24] using artificial anomaly injection in the normal dataset to produce labeled training data set.…”

Section: A Supervised Anomaly Detectionmentioning

confidence: 99%

Adaptive Hybrid Model for Network Intrusion Detection and Comparison among Machine Learning Algorithms

Haque¹,

Al-Kharobi²

2015

IJMLC

View full text Add to dashboard Cite

Abstract-In this paper, we propose a novel method using ensemble learning scheme for classifying network intrusion detection from the most renowned KDD cup dataset. We have shown that reducing the dimensionality of the large dataset provides most accurate detection. Additionally, several machine learning algorithms are used to generate the accuracy metrics and analyzed further for proper comparison. Our approach found out that this algorithm outperforms all other learning techniques. Our goal is to analyze the network intrusion data and find out the best components and use them for the attack analysis. This scheme can be used in parallel with the intrusion detection system to augment its prediction performance for the future data packets. Empirical results show that the input dimensionality reduction can provide lightweight intrusion detection system that can be embedded with the vulnerable system for generating correct classification with significance improvement in execution time.

show abstract

“…This problem in machine learning is known as novelty detection. Furthermore, supervised learning algorithms have been used on synthetic data sets to prove the performance of outlier detection [1,16]. More particularly in retail stores the use of pattern discovery to address retail fraud has been used by Gabbur et al [9] with many limitations to train a classifier due to the lack of reliable fraud data.…”

Section: Related Workmentioning

confidence: 99%

Using the RetSim Fraud Simulation Tool to Set Thresholds for Triage of Retail Fraud

Lopez-Rojas

Axelsson

2015

Secure IT Systems

View full text Add to dashboard Cite

Abstract. The investigation of fraud in business has been a staple for the digital forensics practitioner since the introduction of computers in business. Much of this fraud takes place in the retail industry. When trying to stop losses from insider retail fraud, triage, i.e. the quick identification of sufficiently suspicious behaviour to warrant further investigation, is crucial, given the amount of normal, or insignificant behaviour. It has previously been demonstrated that simple statistical threshold classification is a very successful way to detect fraud [15]. However, in order to do triage successfully the thresholds have to be set correctly. Therefore, we present a method based on simulation to aid the user in accomplishing this, by simulating relevant fraud scenarios that are foreseeing as possible and expected, to calculate optimal threshold limits. Our proposed method gives the advantage over arbitrary thresholds that it reduces the amount of labour needed on false positives and gives additional information, such as the total cost of a specific modelled fraud behaviour, to set up a proper triage process. With our method we argue that we contribute to the allocation of resources for further investigations by optimizing the thresholds for triage and estimating the possible total cost of fraud. Using this method we manage to keep the losses below a desired percentage of sales, which the manager considers acceptable for keeping the business properly running.

show abstract

Outlier detection by active learning

Cited by 260 publications

References 12 publications

Comparative Analysis of Outlier Detection Techniques

Comparative Analysis of Outlier Detection Techniques

Adaptive Hybrid Model for Network Intrusion Detection and Comparison among Machine Learning Algorithms

Using the RetSim Fraud Simulation Tool to Set Thresholds for Triage of Retail Fraud

Contact Info

Product

Resources

About