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...

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...

Android smart phones have become more and popular due to its increasing functionalities, compatibility convenience. More applications been developed can be downloaded easily from app markets. However, malwares increased significantly in recent years. In this paper, we proposed a feature set containing the permissions API calls for malware static detection. Classifiers that used outperform those only with experimentally. It showed information of is helpful recognizing malware.

Face recognition technique has been widely applied to personal identification systems due its satisfying performance. However, security may be a crucial issue, since many studies have shown that face vulnerable in an adversarial environment, which adversary can camouflage as legitimate user order mislead the system. Although liveness detection methods proposed distinguish real and fake faces, they are either time-consuming, costly, or sensitive noise illumination. This paper proposes method...

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,...

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...

The dramatic decline in oil prices, from around $110 per barrel June 2014 to less than $40 March 2016, highlights the importance of competition between different energy sources. Indeed, sustained price drop has been primarily attributed OPEC's strategic decision not curb its production face increased supply shale U.S., spurred by technological innovation "fracking." We study how continuous time Cournot competitions, which firms producing similar goods compete with one another setting...

The short term load forecasting plays an essential role in the operation of electric power systems. Plenty features involved cause a complex system and long training time. curse dimensionality also downgrades generalization capability predictor. This paper applies random forest based ensemble to application. Rather than selecting subset features, which may information lost, all are considered proposed method. Different feature sets used construct regression systems average method is as...

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...

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...