Pattern recognition and machine learning techniques have been increasingly adopted in adversarial settings such as spam, intrusion, malware detection, although their security against well-crafted attacks that aim to evade detection by manipulating data at test time has not yet thoroughly assessed. While previous work mainly focused on devising adversary-aware classification algorithms counter evasion attempts, only few authors considered the impact of using reduced feature sets classifier...

Undersampling is a widely adopted method to deal with imbalance pattern classification problems. Current methods mainly depend on either random resampling the majority class or at decision boundary. Random-based undersampling fails take into consideration informative samples in data while boundary sensitive overlapping. Both techniques ignore distribution information of training dataset. In this paper, we propose diversified sensitivity-based method. Samples are clustered capture and enhance...

Aim to currently content-based image retrieval method having high computational complexity and low accuracy problem, this paper proposes a based on color texture features. As its features, moments of the Hue, Saturation Value (HSV) component images in HSV space are used. Gabor descriptors adopted. Users assign weights each feature respectively calculate similarity with combined features according normalized Euclidean distance. Experiment results show that proposed has higher than...

The training of a multilayer perceptron neural network (MLPNN) concerns the selection its architecture and connection weights via minimization both error penalty term. Different terms have been proposed to control smoothness MLPNN for better generalization capability. However, controlling using, instance, norm or Vapnik-Chervonenkis dimension cannot distinguish individual MLPNNs with same number free parameters norm. In this paper, enhance capabilities, we propose stochastic sensitivity...

In a modern e-commerce recommender system, it is important to understand the relationships among products. Recognizing product relationships-such as complements or substitutes-accurately an essential task for generating better recommendation results, well improving explainability in recommendation. Products and their associated naturally form graph, yet existing efforts do not fully exploit graph's topological structure. They usually only consider information from directly connected fact,...

Radio Frequency Identification (RFID) is the next generation wireless communication technology applicable to a wide range of application areas. There are an increasing number retailers, banks, traffic managements, exhibitions and logistic providers practicing this new their products services. Therefore, it brings both opportunities challenges RFID researchers. In paper, we provide brief survey on applications suggest some in intelligent applications.

This paper presents a covariance-matrix modeling and detection approach to detecting various flooding attacks. Based on the investigation of correlativity changes monitored network features during attacks, this employs statistical covariance matrices build norm profile normal activities in information systems directly utilizes detect The classification boundary is constrained by threshold matrix, where each element evaluates degree which an observed matrix different from terms correlation...

The one-class classification problem aims to distinguish a target class from outliers. spherical classifier (SOCC) solves this by finding hypersphere with minimum volume that contains the data while keeping outlier samples outside. SOCC achieves satisfactory performance only when have same distribution tendency in all orientations. Therefore, of is limited way many superfluous outliers might be mistakenly enclosed. authors propose exploit structures obtained via unsupervised methods such as...

This paper addresses a problem in the hashing technique for large scale image retrieval: learn compact hash code to reduce storage cost with performance comparable that of long code. A longer yields better precision rate retrieved images. However, it also requires larger storage, which limits number stored Current methods employ same length both queries and We propose new scheme using two codes different lengths images, i.e., asymmetric cyclical hashing. is used requirement, while query...

Convolutional Neural Network (CNN) achieved satisfying performance in click-through rate (CTR) prediction recent studies. Since features used CTR have no meaningful sequence nature, the can be arranged any order. As CNN learns local information of a sample, feature may influence its significantly. However, this problem has not been fully investigated. This paper firstly investigates whether and how affects CNN-based method. data distribution changes with time, best current suitable for...

A recommender system is susceptible to manipulation through the injection of carefully crafted profiles. Some recent profile identification methods only perform well in specific attack scenarios. general detection method usually complicated or requires label samples. Such are prone overtraining easily, and process annotation incurs high expenses. This study proposes an unsupervised divide-and-conquer aiming identify profiles, utilizing a specifically designed model for each kind shilling...

A novel adaptive filter is proposed using a recurrent cerebellar-model-articulation-controller (CMAC). The locally globally feedforward CMAC (RCMAC) has favorable properties of small size, good generalization, rapid learning, and dynamic response, thus it more suitable for high-speed signal processing. To provide fast training, an efficient parameter learning algorithm based on the normalized gradient descent method presented, in which rates are on-line adapted. Then Lyapunov function...

Fundus images have been widely used in the diagnosis of retinopathy and cardiovascular diseases. However, because movement patients' eyes limitation medical equipments, quality fundus may be low sometimes. In this paper, we propose vessel enhancement method for a low-contrast blurred image based on multi-scale morphological top-hat transformation, combination Gabor matched filter. The underlying rationale proposed is to make use different kinds information improve blood vessels retinal with...

A very large volume of images is uploaded to the Internet daily. However, current hashing methods for image retrieval are designed static databases only. They fail consider fact that distribution can change when new added database over time. The changes in include both discovery a class and within owing concept drift. Retraining hash tables using all requires computation effort. This also biased old data huge which leads poor performance In this paper, we propose incremental (ICH) method...

Radio frequency identification (RFID) technology has been widely adopted in access control system. However, the people holding RFID card passing through may not be authorized one. Therefore, an system combining and face recognition based on neural network is presented this work. The recognizes of person denies if they do match. We adopt a Radial Basis Function Neural Network (RBFNN) to learn holders save parameters RBFNN only. This could reduce storage when number getting large. Principal...