23 April 2018 in IEEE Access
Collaborative filtering recommendation with implicit feedbacks (e.g., clicks, views, and plays) is regarded as one of the most challenging issues in both the academic and industry. From implicit feedbacks, we can only get a small fraction of observed data (positive examples), and the massive unobserved data are the mixture of negative examples and unlabeled positive examples. However, most of existing efforts either treat unobserved data equally by assigning a uniform weight, or uniformly weight observed data while ignoring the hidden information (i.e., visit frequency) in implicit feedbacks. This assumption may not hold in real-life scenarios since they cannot distinguish the contributions of the whole data, and easily leads to prediction bias. Besides, those approaches still suffer from low-efficiency issue. To this end, we propose an all-weighted matrix factorization and fast optimization strategy for effective and efficient recommendation. We first design a frequency-aware weighting scheme for observed data and a user-oriented weighting scheme for unobserved data non-uniformly. Then, the weighting scheme of both observed and unobserved data are combined in a unified way to form an all-weighted matrix factorization model. Afterwards, we present a surrogate objective function and develop a fast optimization strategy to enhance the efficiency. Extensive experimental results on real-world datasets demonstrate that our method outperforms the competitive baselines on several evaluation metrics.
16 May 2019 in IEEE Access
A serial support vector machine (SVM) ensemble classifier based on feature selection is designed to improve the efficiency and accuracy of unbalanced data classification so as to fully excavate the potential of features. The performance measurement indicators of the traditional SVM model “oneversus- others” are defined. With the goal of reaching higher classification precision and smaller feature set scale, the serial ensemble classifier is built using “one-versus-others” binary SVM classifier as base classifier, ant colony optimization (ACO) is used to determine the recognizable category and input feature subset of each level. Several experiments on vibration signals of real states of engine are performed to contrast the classifying effects of proposed approach, traditional “one-versus-one” and “one-versus-others” parallel SVM classifiers, results demonstrate the superiority of proposed approach.
8 May 2019 in IEEE Transactions on Neural Networks and Learning Systems
Dissimilarity measures play a crucial role in clustering and, are directly related to the performance of clustering algorithms. However, effectively measuring the dissimilarity is not easy, especially for categorical data. The main difficulty of the dissimilarity measurement for categorical data is that its representation lacks a clear space structure. Therefore, the space structure-based representation has been proposed to provide the categorical data with a clear linear representation space. This representation improves the clustering performance obviously but only applies to small data sets because its dimensionality increases rapidly with the size of the data set. In this paper, we investigate the possibility of reducing the dimensionality of the space structure-based representation while maintaining the same representation ability. A lightweight representation scheme is proposed by taking a set of representative objects as the reference system (called the reference set) to position other objects in the Euclidean space. Moreover, a preclustering-based strategy is designed to select an appropriate reference set quickly. Finally, the representation scheme together with the $k$ -means algorithm provides an efficient method to cluster the categorical data. The theoretical and the experimental analysis shows that the proposed method outperforms state-of-the-art methods in terms of both accuracy and efficiency.
28 January 2022 in Security and Communication Networks
Attacker identification from network traffic is a common practice of cyberspace security management. However, network administrators cannot cover all security equipment due to the cyberspace management cost constraints, giving attackers the chance to escape from the surveillance of network security administrators by legitimate actions and to perform the attack in both physical domain and digital domain. Therefore, we proposed a hidden attack sequence detection method based on reinforcement learning to deal with the challenge through modeling the network administrators as an intelligent agent that learns their action policy from the interaction with the cyberspace environment. Following Deep Deterministic Policy Gradient (DDPG), the intelligent agent can not only discover the hidden attackers hiding in the legitimate action sequences but also reduce the cyberspace management cost. Furthermore, a dynamic reward DDPG method was proposed to improve defense performance, which set dynamic reward depending on the hidden attack sequences steps and agent’s check steps, compared to the fixed reward in common methods. Meanwhile, the method was verified in a simulated experimental cyberspace environment. Finally, the experimental results demonstrate that there are hidden attack sequences in cyberspace, and the proposed method can discover the hidden attack sequences. The dynamic reward DDPG shows superior performance in detecting hidden attackers, with a detection rate of 97.46%, which can improve the ability to discover hidden attackers and reduce the 6% cyberspace management cost compared to DDPG.
12 August 2019 in IEEE Access
Truth discovery methods aim to identify which piece of information is trustworthy from multi-sourced data. Most existing truth discovery methods, however, are designed for structured data and fail to meet the strong need to extract trustworthy information from raw text data. More specifically, existing methods ignore the semantic information of text answers, i.e., answers may contain multiple factors, the word usages may be diverse, and the answers may be partially correct. In addition, ubiquitous long-tail phenomenon exists in the tasks, i.e., most users provide only a few answers and only a few users provide plenty of answers, which causes the user reliability estimation for small users to be unreasonable. To tackle these challenges, we propose a Graph Convolutional Network (GCN) based truth discovery model to automatically discover trustworthy information from text data. Firstly, Smooth Inverse Frequency (SIF) is utilized to learn real-valued vector representations for answers. Then, we construct undirected graph with these vectors to capture the structural information of answers. Finally, the GCN is utilized to store and update the reliability of these answers, and sums up all the feature vectors of all neighboring answers to improve the accuracy and efficiency of truth discovery. Different from traditional methods, we use vectors to store the reliability of answers which have higher representation capability compared with real numbers, and network is utilized to capture complex relationships among answers rather than simplified functions. The experiment results on real datasets show that though text data structures are complex, our model can still find reliable answers compared with retrieval-based and state-of-the-art truth discovery methods.
10 October 2022 in Scientific Programming
There are certain vulnerabilities at the beginning of multiple domain cyberspace configuration. How to discover these potential vulnerabilities has been a hot topic. This paper proposes to find these cyberspace vulnerabilities by discovering the shortest attack path in multiple domain cyberspace. In order to discover more and shorter attack paths, we train an agent as an attacker to discover multiple domain cyberspace attack paths. We formulate the discovering attack paths as a reinforcement learning (RL) problem. With this technique, we added a multiple domain action select module to RL that can pick an executable action in a state. By using the proposed method, we can discover more hidden attack paths and shorter attack paths to analyze the potential vulnerabilities to cyberspace. Finally, we created a simulated cyberspace experimental environment to test our proposed method. The experimental results show that the proposed method can discover more hidden multiple domain attack paths and shorter attack paths than the existing methods.