2015
DOI: 10.1049/iet-ifs.2014.0488
|View full text |Cite
|
Sign up to set email alerts
|

Robust collaborative recommendation algorithm based on kernel function and Welsch reweighted M‐estimator

Abstract: The existing collaborative recommendation algorithms based on matrix factorisation (MF) have poor robustness against shilling attacks. To address this problem, in this study the authors propose a robust collaborative recommendation algorithm based on kernel function and Welsch reweighted M-estimator. They first propose a median-based method to calculate user and item biases, which can reduce the influence of shilling attacks on user and item biases because median is insensitive to outliers. Then, they present … Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
4
1

Citation Types

0
9
0

Year Published

2017
2017
2021
2021

Publication Types

Select...
4
2
1

Relationship

0
7

Authors

Journals

citations
Cited by 9 publications
(9 citation statements)
references
References 38 publications
0
9
0
Order By: Relevance
“…a model of privacy conservation better fight six shilling attacks by dismantling shilling attack data and then K-means, a discrete transformation of the wavelet, a singular value decomposition and itembased prediction algorithms. This approach offers greater robustness in model-based schemes against shilling.a recommended model to provide improved rating identification than matrix-factorized IoT networks has been developed for the kernel with Welsch weightedm-estimator (Zhang, F., et al 2015). The user rating is calculated using a median formula that compares the user rating with the object rating.…”
Section: Related Workmentioning
confidence: 99%
See 1 more Smart Citation
“…a model of privacy conservation better fight six shilling attacks by dismantling shilling attack data and then K-means, a discrete transformation of the wavelet, a singular value decomposition and itembased prediction algorithms. This approach offers greater robustness in model-based schemes against shilling.a recommended model to provide improved rating identification than matrix-factorized IoT networks has been developed for the kernel with Welsch weightedm-estimator (Zhang, F., et al 2015). The user rating is calculated using a median formula that compares the user rating with the object rating.…”
Section: Related Workmentioning
confidence: 99%
“…during collective filtering, a number of different new algorithms discovered the shilling attack successfully (bilge, a., et al 2014 andZhang, F., et al 2015). These algorithms are meant to enhance the algorithms that provide robustness with basic truth knowledge against shillers.…”
Section: Introductionmentioning
confidence: 99%
“…Unfortunately, they happen more in reality than we hoped. Up to now, many types the cost functions have been used to design adaptive filtering algorithms, such as least mean absoult third 18 , least mean fourth 19 , entropy 20 , least-squares estimator 21 , absolute value estimator 22 , the Cauchy 23 , Geman–McClure 24 , Welsch 25 , 26 , and Huber function 27 30 . Specifically, the least-squares estimator is not robust because their influence function is not bounded 21 , and the absolute value estimator is not stable because the function of estimate error ( ) at i -th is not strictly convex 22 .…”
Section: Introductionmentioning
confidence: 99%
“…As can be seen from the influence function, the influence of large estimation errors only decreases linearly with their size. The Geman–McClure 24 and Welsch 25 , 26 functions try to further reduce the effect of large estimation errors. It seems complicated to select a cost function for general use without being somewhat arbitrary.…”
Section: Introductionmentioning
confidence: 99%
“…In order to improve the attack-resistant ability of collaborative recommender systems, researchers have proposed many shilling attack detection algorithms [4][5][6] and robust recommendation algorithms [7][8][9][10]. Mehta et al proposed shilling attack detection algorithm using principal component analysis (PCA) of the users' profiles [4].…”
Section: Introductionmentioning
confidence: 99%