Massive dissemination of fake news and its potential to erode democracy has increased the demand for accurate detection. Recent advancements in this area have proposed novel techniques that aim detect by exploring how it propagates on social networks. Nevertheless, at an early stage, i.e., when is published a outlet but not yet spread media, one cannot rely propagation information as does exist. Hence, there strong need develop approaches can focusing content. In article, theory-driven model...

In this paper, we present "k-means+ID3", a method to cascade k-means clustering and the ID3 decision tree learning methods for classifying anomalous normal activities in computer network, an active electronic circuit, mechanical mass-beam system. The first partitions training instances into k clusters using Euclidean distance similarity. On each cluster, representing density region of or anomaly instances, build tree. on cluster refines boundaries by subgroups within cluster. To obtain final...

A chipless RF identifiction (RFID) sensor system platform consisting of passive RFID tags and specialized reader has been developed for cyber centric monitoring applications. The are fabricated on a flexible substrate, the tag identification (ID) generation circuit consists microstrip transmission lines. Two configurations presented. first configuration (conf-I) an antenna integrated sensor. second (conf-II) antenna, ID circuit, system, analog section on-board computer communicates with...

Research on attacks which exploit video-based side-channels to decode text typed a smartphone has traditionally assumed that the adversary is able leverage some information from screen display (say, reflection of or low resolution video content screen). This paper introduces new breed side-channel attack PIN entry process entirely relies spatio-temporal dynamics hands during typing text. Implemented dataset 200 videos an HTC One phone, we show, breaks average over 50% PINs first attempt and...

While body movement patterns recorded by a smartphone accelerometer are now well understood to be discriminative enough separate users, little work has been done address the question of if or how position in which phone is held affects user authentication. In this work, we show through combination supervised learning methods and statistical tests, that there certain users for whom exploitation information drastically improves classification performance. We propose two-stage authentication...

Despite the tremendous need for evaluation of touch-based authentication as an extra security layer mobile devices, huge disparity in experimental methodology used by different researchers makes it hard to determine how much research this area has progressed. Critical variables such types features and they are pre-processed, training testing performance metrics, mention but a few, vary from one study next. Additionally, most datasets these evaluations not openly accessible, making impossible...

Smart wearable devices have recently become one of the major technological trends and been widely adopted by general public. Wireless earphones, in particular, seen a skyrocketing growth due to its great usability convenience. With goal seeking more unobtrusive authentication method that users can easily use conveniently access, this study we present EarEcho as novel, affordable, user-friendly biometric solution. takes advantages unique physical geometrical characteristics human ear canal...

Touch-based verification --- the use of touch gestures (e.g., swiping, zooming, etc.) to authenticate users screen devices has recently been widely evaluated for its potential serve as a second layer defense PIN lock mechanism. In all performance evaluations touch-based authentication systems however, researchers have assumed naive (zero-effort) forgeries in which attacker makes no effort mimic given gesture pattern.

We studied the fusion of three biometric authentication modalities, namely, swiping gestures, typing patterns and phone movement observed during or swiping. A web browser was customized to collect data generated from aforementioned modalities over four seven days in an unconstrained environment. Several features were extracted by using sliding window mechanism for each modality analyzed information gain, correlation, symmetric uncertainty. Finally, five windows continuous swipes, thirty...

3D printing, or additive manufacturing, is a key technology for future manufacturing systems. However, printing systems have unique vulnerabilities presented by the ability to affect infill without affecting exterior. In order detect malicious defects in process, this paper proposes following: 1) investigate 2) extract features based on simulated process images, and 3) an experiment of image classification with one group non-defect other defect training from process. The images are captured...

We propose a Monte Carlo approach to attain sufficient training data, splitting method improve effectiveness, and system composed of parallel decision trees (DTs) authenticate users based on keystroke patterns. For each user, approximately 19 times as much simulated data was generated complement the 387 vectors raw data. The set, including is split into four subsets. subset, wavelet transforms are performed obtain total eight subsets for user. Eight DTs thus trained using A DT constructed...

Hidden Markov Model (HMM) based applications are common in various areas, but the incorporation of HMM's for anomaly detection is still its infancy. This paper aims at classifying TCP network traffic as an attack or normal using HMM. The paper's main objective to build system, a predictive model capable discriminating between and abnormal behavior traffic. In training phase, special attention given initialization selection issues, which makes phase particularly effective. For HMM, 12.195%...

We present a new attack called the snoop-forge-replay on keystroke-based continuous verification systems. The is sample-level forgery and not specific to any particular method or system. It can be launched with easily available keyloggers APIs for keystroke synthesis. Our results from 2640 experiments show that: 1) attacks achieve alarmingly high error rates compared zero-effort impostor attacks, which have been de facto standard evaluating systems; 2) four state-of-the-art methods, three...

Research on keystroke-based authentication has traditionally assumed human impostors who generate forgeries by physically typing the keyboard. With bots now well understood to have capacity originate precisely timed keystroke sequences, this model of attack is likely underestimate threat facing a system in practice. In work, we investigate how would perform if it were subjected synthetic attacks designed mimic typical user. To implement attacks, rigorous statistical analysis biometrics data...

We derive the feature selection criterion presented in [CHECK END OF SENTENCE] and from multidimensional mutual information between features class. Our derivation: 1) specifies validates lower-order dependency assumptions of 2) mathematically justifies utility by relating it to Bayes classification error.

In this paper, we demonstrate that gait patterns of an individual captured through a smartphone accelerometer can be imitated with the support digital treadmill. Furthermore, design attack for baseline based authentication system (GBAS) and rigorously test its performance over eighteen user data-set. By employing only two imitators using simple treadmill speed control functionality, increases average false acceptance rate (FAR) from 5.8% to 43.66% random forest, best performing classifier in...

Individuals can be misled by fake news and spread it unintentionally without knowing is false. This phenomenon has been frequently observed but not investigated. Our aim in this work to assess the intent of spreaders. To distinguish between intentional versus unintentional spreading, we study psychological explanations spreading. With foundation, then propose an influence graph, using which extensive experiments show that assessed help significantly differentiate Furthermore, estimated...

In this work, we present a novel approach to clustering Web site users into different groups and generating common user profiles. These profiles can be used make recommendations, personalize sites, for other uses such as targeting advertising. By using the concept of mass distribution in Dempster-Shafer's theory, belief function similarity measure our algorithm adds task ability capture uncertainty among user's navigation behavior. Our is relatively simple use gives comparable results...

The degree of personalization that a Web site offers in presenting its services to users is an important attribute contributing the site's popularity. server access logs contain substantial data about user patterns. One way solve this problem group on basis their interests and then organize structure according needs different groups. Two main difficulties inhibit approach: essentially infinite diversity change these with time. We have developed clustering algorithm groups based ART1 version...