/sequential-attack-detector

Python implementation of sequential attack detector for recommender systems

Primary LanguagePython

Sequential Attack Detector

This code is Python implementation of the paper "Sequential Attack Detection in Recommender Systems".

Model

The framework consists of a latent variable model, which is trained given the rating data and user/item attributes, and a CUSUM-like sequential detector to test newly registered users to detect shilling attacks by exploiting the uni-variate statistics from the latent space.

Dataset

Movielens 1M dataset is provided for a demonstration.