DeepScan: Exploiting Deep Learning for Malicious Account Detection in Location-Based Social Networks

  title={DeepScan: Exploiting Deep Learning for Malicious Account Detection in Location-Based Social Networks},
  author={Qingyuan Gong and Yang Chen and Xinlei He and Zhou Zhuang and Tianyi Wang and Hong Huang and Xin Wang and Xiaoming Fu},
  journal={IEEE Communications Magazine},
Our daily lives have been immersed in widespread location-based social networks (LBSNs. [] Key Method Different from existing approaches, DeepScan leverages emerging deep learning technologies to learn users' dynamic behavior. In particular, we introduce the long short-term memory (LSTM) neural network to conduct time series analysis of user activities.

Figures and Tables from this paper

Deep Learning-Based Malicious Account Detection in the Momo Social Network
This paper explores the malicious account detection problem by introducing a deep learning-based framework using the long short-term memory (LSTM) neural network, and is able to build a classifier to achieve the binary classification.
DeepFriend: finding abnormal nodes in online social networks using dynamic deep learning
A model to classify malicious vertices using nodes' link information by training extensive features with dynamic deep learning architecture is proposed and gains higher accuracy than standard learning algorithms in the abnormal nodes’ classification.
Detecting Malicious Accounts in Online Developer Communities Using Deep Learning
This work formulated the malicious account detection problem in online developer communities, and proposed GitSec, a deep learning-based solution to detect malicious accounts, and showed that GitSec is an accurate detection system.
Automatic ICA detection in online social networks with PageRank
An automatic method to identify cloned profiles is proposed and is implemented on Hadoop framework using the MapReduce programming model, which employs a distributed processing framework, limits the search space, and decreases the required computation by clustering the profiles.
Cross-site Prediction on Social Influence for Cold-start Users in Online Social Networks
This work proposes a practical solution to predict whether a cold-start user will become an influential user on an emerging OSN, by opportunistically leveraging the user’s information on dominant OSNs, and achieves a high prediction performance based on different social influence definitions.
A Feature Based Approach to Detect Fake Profiles in Twitter
This work aims to use a feature-based approach to identify fake profiles on social media platforms and shows that the proposed approach is efficient in detecting fake profiles.
Classification of various attacks and their defence mechanism in online social networks: a survey
A high-level classification of recent OSN attacks for recognising the problem and analysing the blow of such attacks on World Wide Web is presented and some simple-to-implement user practice tips to protect the system and user’s information are offered.
Identification of Influential Users in Emerging Online Social Networks Using Cross-site Linking
A supervised machine learning-based system is built by leveraging the widely adopted cross-site linking function, which could overcome the limitations of referring to the user data of a single OSN.
Twitter Fake Account Detection and Classification using Ontological Engineering and Semantic Web Rule Language
A new approach with dual functions, namely to identify and classify the twitter bots based on ontological engineering and Semantic Web Rule Language (SWRL) rules, and it has been found that he ontology classifier is a more interpretable model that offers straightforward and human-interpretable decision rules, as compared to other machine learning classifiers.
Online Social Networks Misuse, Cyber Crimes, and Counter Mechanisms
This chapter presents an overview of various cyber-crimes associated with OSN environment to gain insight into ongoing cyber-attacks and counter mechanisms in the form of tools, techniques, and frameworks are suggested.


Uncovering social network Sybils in the wild
These efforts to detect, characterize, and understand Sybil account activity in the Renren Online Social Network (OSN) are described and it is shown that Sybils can effectively avoid existing community-based Sybil detectors.
Aiding the Detection of Fake Accounts in Large Scale Social Online Services
A new tool in the hands of OSN operators, which relies on social graph properties to rank users according to their perceived likelihood of being fake (SybilRank), which is computationally efficient and can scale to graphs with hundreds of millions of nodes, as demonstrated by the Hadoop prototype.
Analyzing and Detecting Opinion Spam on a Large-scale Dataset via Temporal and Spatial Patterns
This work presents the first large-scale analysis of restaurant reviews filtered by Dianping's fake review filtering system, and proposes some novel temporal and spatial features for supervised opinion spam detection that significantly outperform existing state-of-art features.
DeepMood: Forecasting Depressed Mood Based on Self-Reported Histories via Recurrent Neural Networks
Experimental results show that the method forecast the severely depressed mood of a user based on self-reported histories, with higher accuracy than SVM, and that the long-term historical information of a users improves the accuracy of forecasting depressed mood.
We know how you live: exploring the spectrum of urban lifestyles
An algorithm to connect multiple social network accounts of millions of individuals and collect their publicly available heterogeneous behavioral data as well as social links is developed and a nonparametric Bayesian approach is developed to model the lifestyle spectrum of a group of individuals.
Long Short-Term Memory
A novel, efficient, gradient based method called long short-term memory (LSTM) is introduced, which can learn to bridge minimal time lags in excess of 1000 discrete-time steps by enforcing constant error flow through constant error carousels within special units.
XGBoost: A Scalable Tree Boosting System
This paper proposes a novel sparsity-aware algorithm for sparse data and weighted quantile sketch for approximate tree learning and provides insights on cache access patterns, data compression and sharding to build a scalable tree boosting system called XGBoost.
Revealing the City That We Cannot See
The City Image visualization technique is proposed, which builds a transition graph mapping people's movements between location categories, and its use to identify similarities and differences of human dynamics across cities by clustering cities according to their citizens' routines is demonstrated.
A Comparative Study on Feature Selection in Text Categorization
This paper finds strong correlations between the DF IG and CHI values of a term and suggests that DF thresholding the simplest method with the lowest cost in computation can be reliably used instead of IG or CHI when the computation of these measures are too expensive.
Framewise phoneme classification with bidirectional LSTM networks
  • A. GravesJ. Schmidhuber
  • Computer Science
    Proceedings. 2005 IEEE International Joint Conference on Neural Networks, 2005.
  • 2005
It is found that bidirectional LSTM outperforms both RNNs and unidirectionalLSTM, and the significance of framewise phoneme classification to continuous speech recognition and the validity of usingbidirectional networks for online causal tasks is discussed.