Search

Due to Twitter’s potential reach and influence, malicious automated accounts and services have been operating and growing without control. One of the most recognizable is the bot, a piece of programming or automated system reproducing an assignment constantly over time. The bots’ owners can also coordinate them into botnets to increase their capabilities and impact on the platform. In the literature, botnet detection on Twitter is a relatively under-researched topic where researchers need complete and updated collections, as many of the botnets currently reported are outdated or incomplete. Despite Twitter’s best efforts to mitigate the botnets’ presence by issuing restrictions and bans, they are insufficient due to their accelerated creation and expansion. Therefore, we propose a solution employing an updated botnet collection that identifies with high certainty which botnets a set of bots belongs to. Our experimentation showed that our solution labeled the botnets accurately, achieving an outstanding performance with unknown botnets in most of our experiments and when compared with other works from the literature. We also proved that our solution obtained its best scores without relying on a specific algorithm or configuration.

Commercially available quantum computers are expected to reshape the world in the near future. They are said to break conventional cryptographic security mechanisms that are deeply embedded in our today’s communication. Symmetric cryptography, such as AES, will withstand quantum attacks as long as the key sizes are doubled compared to today’s key lengths. Asymmetric cryptographic procedures, e.g. RSA, however are broken. It is therefore necessary to change the way we assure our privacy by adopting and moving towards post-quantum cryptography (PQC) principles. In this work, we benchmark three PQC algorithms, Falcon, Dilithium, and Kyber. Moreover, we present an implementation of a PQC stack consisting of the algorithms Dilithium/Kyber and Falcon/Kyber which use hardware accelerators for some key functions and evaluate their performance and resource utilization. Regarding a classic server-client model, the computational load of the Dilithium/Kyber stack is distributed more equally among server and client. The stack Falcon/Kyber biases the computational challenges towards the server, hence relieving the client of performing costly operations. We found that Dilithium’s advantage over Falcon is that Dilithium’s execution is faster while the workload to be performed is distributed equally among client and server, whereas Falcon’s advantage over Dilithium lies within the small signature sizes and the unequally distributed computational tasks. In a client server model with a performance limited client (i.e. Internet-of-Things - IoT - environments) Falcon could proof useful for it constrains the computational hard tasks to the server and leaves a minimal workload to the client. Furthermore, Falcon requires smaller bandwidth, making it a strong candidate for deep-edge or IoT applications.

Microtransactions, small in-app purchases within virtual environments, have become increasingly prevalent in modern gaming, often leading to addictive behaviors and financial burdens, particularly among young male players. The potential consequences of such microtransactions within the metaverse extend beyond gaming, as they provide companies with unprecedented control over user experiences and the ability to capitalize personal data for targeted advertising and manipulative practices. This article presents a comprehensive framework for the ethical analysis and design of microtransactions in the metaverse to address these concerns. This three-dimensional framework explores Motivation, Engagement, and Fairness to evaluate the impact of microtransactions on user autonomy, social interactions, and overall fairness. Additionally, the paper conducts a running example on World of Warcraft to assess the alignment of its microtransaction design choices with the proposed ethical principles. By shedding light on these issues, this article aims to raise awareness among consumers, developers, and policymakers about the need for responsible practices, consumer protection, and privacy preservation in the metaverse. It emphasizes fostering collaboration and transparency to safeguard users from misuse while promoting a user-centric metaverse ecosystem.

When facing a classification problem, data science practitioners must search through an armory of methods. Often, practitioners are tempted to use off-the-shelf classifiers, including automated Machine Learning (AutoML) toolboxes; however, stand-alone classifiers are not applicable to every problem and AutoML may be time-consuming raising up environment-ethical issues. To magnify the problem, (commercial) AutoML toolboxes are black and practitioners are not allowed to extend them with new methods to improve their classification performance. Our main objective is to show that an off-the-shelf classifier designed for class imbalance problems can achieve similar performance to an AutoML toolbox. To do so, first, we present the student dropout prediction case study, which most off-the-shelf classifiers find difficult to solve due to the problem’s inherent class imbalance. We show that Microsoft Azure AutoML outperforms several popular, stand-alone classifiers. However, multivariate PBC4cip, an off-the-shelf classifier especially designed to deal with class imbalance, yields results that are just as good as Microsoft Azure AutoML, with the advantage that the expensive steps of mechanism selection and tuning are avoided. Our studies show that data science practitioners need to build themselves a taxonomy of classification mechanisms in terms of the properties of the problem to solve. Additionally, AutoML platforms should let scientists modify the armory of classifiers and provide an explanation of both mechanism selection and mechanism tunning so that practitioners learn further lessons.

In this paper, we propose a novel hybrid network solution using optical code-division multiple-access and free-space optics (OCDMA-FSO) to support multiple data rates and quality-of-service according to user requirements. The proposed OCDMA-FSO network classifies users into diverse classes based on their requirements and allows for signal transmission in a fully asynchronous manner. The new hybrid network employs 2-D OCDMA codes as well as a variable aperture FSO receiver together with an equal gain combining (EGC) technique. We derive completely new analytical formalisms to evaluate the network performance in terms of average bit error rate (ABER), outage probability, effective data rate during outages, and ergodic capacity. The derived ABER expressions are validated via Monte-Carlo simulations. These new analytical formalisms are derived considering several atmospheric turbulence conditions such as uncorrelated, totally correlated, and partially correlated turbulence. To model this atmospheric turbulence, we derive the gamma-gamma (GG) distribution, which allows us to consider both the scintillation effect and the EGC technique. Finally, we carry out a comprehensive performance evaluation of the proposed hybrid network. Analytical results show that as the number of FSO receiver apertures increases, there is an improvement in overall performance levels. However, the ergodic capacity exhibits minimal improvement even as the number of FSO receiver apertures increases, regardless of the user requirements and turbulence conditions. The proposed network is a potential solution for scenarios that demand flexibility in bandwidth and data rate allocation.

Insurance Telematics is a recent service that is offered to drivers where a sensing platform measures vehicle’s dynamics with the ultimate goal of inferring the level of risk that the driver experiences on the road. For this to happen, a robust set of sensors must be deployed within the vehicle and effective algorithms to process and classify driving patterns must be employed. As feasible as it looks, however, insurance telematics poses a great challenge when sensors offer information that is redundant or even show a low signal/noise ratio. The question here, then, is how to identify what sensing information is really useful and discard that one that only adds overhead. In this study we proposed an algorithmic approach, called minimal Gated Multimodal Unit (mGMU) that learns to filter out the proper amount of data from different input sources, making it an appealing choice for data fusion scenarios. We evaluate mGMU on three public multi modal (multiple sensor) datasets, created to classify driving styles, that comprise 3, 3 and 8 sensor sources. After a very extensive experimental setup that comprised around 30 different pipelines (made of feature extractors, information fusion approaches and a classification layer), results show that configurations based on mGMU obtain the best performance (with an increase of 2% with respect the second best approach and even outperforming the use of AutoEncoders by 5%). Moreover, as a proof of concept we deployed mGMU on three commercial smartphones, thus showing its efficiency given the number of parameters and FLOPS that were dramatically reduced. Given these results, we expect mGMU to be of great use on resource-constrained edge computing devices and rich sensor scenarios where data from different data sources must be exploited.

This paper presents a 7-parameter analytical model of the MOS transistor based on the inversion charge targeted at the development of simplified analytical circuit design methodologies that take into account the physics of the MOS transistor. The proposed design-oriented model allows for the first time to describe both the main short-channel effects of advanced nanometric technologies and the dependence of the transistor drain current on the drain voltage, while the model remains valid for all bias regimes (from weak to strong inversion) and for all operating regions (linear and saturated). A simple procedure based on the device physics is proposed to estimate the transistor model parameters for a given technology. Furthermore, analytical expressions of the current derivatives are developed targeting different design scenarios. The accuracy of the proposed model is validated by direct comparison to silicon measurements of N-MOS transistors in 28 nm FD-SOI technology for channel width of $1~\mu \text{m}$ and channel lengths of 30 nm, 60 nm and 150 nm, and also to simulations performed with an industry-standard compact model.

Decision trees are popular as stand-alone classifiers or as base learners in ensemble classifiers. Mostly, this is due to decision trees having the advantage of being easy to explain. To improve the classification performance of decision trees, some authors have used Multivariate Decision Trees (MDTs), which allow combinations of features when splitting a node. While there is growing interest in the area, recent research in MDTs all have in common that they do not provide adequate comparison of related work: they do not consider relevant rival techniques, or they test algorithm performance in an insufficient number of databases. As a result, claims have no statistical sustain and, hence, there is a lack of general understanding of the actual capabilities of existing MDT induction algorithms, crucial to improving the state-of-the-art. In this paper, we report on an exhaustive review of MDTs. In particular, we give an overview of 37 MDT induction algorithms, out of which we have experimentally compared 19 of them in 57 databases. We provide a statistical comparison in all databases and subsets of databases according to the number of classes, number of features, number of instances, and degree of class imbalance. This allows us to identify groups of top-performing algorithms for different types of databases.

The Video game-Crowdsourcing model to recollect data motivates people to participate by entertaining them. Research showed that the solutions players make in this model are competitive against experts in the area. Yet, the studies in the area focus on mimicking people’s behavior, including their mistakes. Therefore, we use a Video game-Crowdsourcing to model a problem of interest to find strategies for it. To describe matches from the video game we created, we designed a representation that simplifies the discovery of strategies. Our experimentation compares high score matches against low score ones to find the best behaviors. We played 13 matches employing a known strategy for the problem to validate the methodology. Then, we applied the methodology to matches from players. The results suggest that extracting sub-sequences is a process to find strategies and that we can use them to design algorithms to improve current algorithmic solutions for that problem.

Latent fingerprint identification is one of the leading forensic activities to clarify criminal acts. However, its computational cost hinders the rapid decision making in the identification of an individual when large databases are involved. To reduce the search time used to generate the fingerprint candidates’ order to be compared, fingerprint indexing algorithms that reduce the search space while minimizing the increase in the error rate (compared to the identification) are developed. In the present research, we propose an algorithm for indexing latent fingerprints based on minutia cylinder codes (MCC). This type of minutiae descriptor presents a fixed structure, which brings advantages in terms of efficiency. Besides, in recent studies, this descriptor has shown an identification error rate, at the local level, lower than the other descriptors reported in the literature. Our indexing proposal requires an initial step to construct the indices, in which it uses k-means++ clustering algorithm to create groups of similar minutia cylinder codes corresponding to the impressions of a set of databases. K-means++ allows for a better outcome over other clustering algorithms because of the selection of the proper centroids. The buckets associated with each index are populated with the background databases. Then, given a latent fingerprint, the algorithm extracts the minutia cylinder codes associated with the clusters’ indices with the lowest distance respect to each descriptor of this latent fingerprint. Finally, it integrates the votes represented by the fingerprints obtained to select the candidate impressions. We conduct a set of experiments in which our proposal outperforms current rival algorithms in presence of different databases and descriptors. Also, the primary experiment reduces the search space by four orders of magnitude when the background database contains more than one million impressions.

Nowadays, the international scientific community of machine learning has an enormous campaign in favor of creating understandable models instead of black-box models. The main reason is that experts in the application area are showing reluctance due to black-box models cannot be understood by them, and consequently, their results are difficult to be explained. In unsupervised problems, where experts have not labeled objects, obtaining an explanation of the results is necessary because specialists in the application area need to understand both the applied model as well as the obtained results for finding the rationale behind each obtained clustering from a practical point of view. Hence, in this paper, we introduce a clustering based on decision trees (eUD3.5), which builds several decision trees from numerical databases. Unlike previous solutions, our proposal takes into account both separation and compactness for evaluating a feature split without decreasing time efficiency and with no empirical parameter to control the depth of the trees. We tested eUD3.5 on 40 numerical databases of UCI Machine Learning Repository, showing that our proposal builds a set of high-quality unsupervised decision trees for clustering, allowing us to obtain the best average ranking compared with other popular state-of-the-art clustering solutions. Also, from the collection of unsupervised decision trees induced by our proposal, a set of high-quality patterns are extracted for showing the main feature-value pairs describing each cluster.

In the above article [1], equations (33) and (36) had a typing error. The correct equations are:

Detecting non-human activity in social networks has become an area of great interest for both industry and academia. In this context, obtaining a high detection accuracy is not the only desired quality; experts in the application domain would also like having an understandable model, with which one may explain a decision. An explanatory decision model may help experts to consider, for example, taking legal action against an account that has displayed offensive behavior, or forewarning an account holder about suspicious activity. In this paper, we shall use a pattern-based classification mechanism to social bot detection, specifically for Twitter. Furthermore, we shall introduce a new feature model for social bot detection, which extends (part of) an existing model with features out of Twitter account usage and tweet content sentiment analysis. From our experimental results, we shall see that our mechanism outperforms other, state-of-the-art classifiers, not based on patterns; and that our feature model yields better classification results than others reported on in the literature.

Latent fingerprint identification is attracting increasing interest because of its important role in law enforcement. Although the use of various fingerprint features might be required for successful latent fingerprint identification, methods based on minutiae are often readily applicable and commonly outperform other methods. However, as many fingerprint feature representations exist, we sought to determine if the selection of feature representation has an impact on the performance of automated fingerprint identification systems. In this paper, we review the most prominent fingerprint feature representations reported in the literature, identify trends in fingerprint feature representation, and observe that representations designed for verification are commonly used in latent fingerprint identification. We aim to evaluate the performance of the most popular fingerprint feature representations over a common latent fingerprint database. Therefore, we introduce and apply a protocol that evaluates minutia descriptors for latent fingerprint identification in terms of the identification rate plotted in the cumulative match characteristic (CMC) curve. From our experiments, we found that all the evaluated minutia descriptors obtained identification rates lower than 10% for Rank-1 and 24% for Rank-100 comparing the minutiae in the database NIST SD27, illustrating the need of new minutia descriptors for latent fingerprint identification.

There is a growing interest in the development of classifiers based on contrast patterns (CPs); partly due to the advantage of them being able to explain classification results in a language that is easy to understand for an expert. CP-based classifiers, when using contrast patterns extracted by miners based on decision trees, attain accuracies comparable with other state-of-the-art classifiers. The existing decision tree-based miners use univariate decision trees (UDTs) to extract CPs. In this paper, we define the concept of multivariate CP. We introduce a multivariate CP miner based on multivariate decision trees (MDTs) as well as a new filtering algorithm for multivariate CPs. From our experimental results, we conclude that our proposed CP miner allows obtaining significantly better classification results than the other state-of-the-art classifiers.

This article presents the development of a dynamic optimization method for controller tuning. This is proposed because most traditional tuning methods for complex coupled dynamic models are based on experience, and thus lacking accuracy. The case study is a Mobile Manipulator that consists of an anthropomorphic manipulator and a differential mobile platform. The system model has a trajectory generator that includes the coupled kinematic model, the Jacobian model, the coupled dynamic model, and a Proportional-Derivative plus controller. The tuning of the model is obtained by solving an optimization problem, using the Differential Evolution algorithm. This optimization approach allows to minimize simultaneously the energy consumption and the error on the trajectory tracking by the end effector. A novel strategy is applied to formulate the objective function, including constant weights for balancing the minimization effect. The objective is to avoid an energy consumption equal to zero that represents an error condition of no motion. The results of the case study and its statistical analysis are presented. The best result was modeled in SolidworksR and simulated in MatlabR. This model was exported to Simscape Multibody™ of Matlab, and its simulation produced satisfactory results, suggesting that the proposed optimization method can be a useful tool to solve real engineering problems.

DNS DDoS attacks may severely affect the operation of computer networks, prompting the need for methods able to timely detect them, and then to apply mitigation countermeasures. Visual models have been used to detect an ongoing DDoS attack, but often demand continuous attention from IT staff. However, machine learning techniques could complement a visual model with further information and with on-time alerts that could help IT officers give attention only when an attack is in progress at its very early stage. In this paper, we present DNS-ADVP, a DNS Anomaly Detection Visual Platform, which, in an integrated manner, provides a novel visualisation that depicts on-line DNS traffic, and a one-class classifier that deals with traffic anomaly detection. Using the visual mode, an IT officer may interpret the current state of traffic for an authoritative DNS server; the model comes with visual semaphores, controlled by the one-class classifier. Due to the highly dynamic nature of DNS traffic, our classification method continuously updates what counts as normal behaviour; it has been successfully tested on synthetic attacks, with an 83% of the area under the curve (AUC). DNS-ADVP is currently in use to real-time monitoring an actual authoritative DNS server.

In this paper, new analytical formalisms to evaluate the packet throughput of multiservice multirate slotted ALOHA optical code-division multiple-access (OCDMA) networks are proposed. The proposed formalisms can be successfully applied to 1-D and 2-D OCDMA networks with any number of user classes in the system. The bit error rate (BER) and packet correct probability expressions are derived, considering the multiple-access interference as binomially distributed. Packet throughput expressions, on the other hand, are derived considering Poisson, binomial, and Markov chain approaches for the composite packet arrivals distributions, with the latter defined as benchmark. A throughput performance evaluation is carried out for two distinct user code sequences separately, namely, 1-D and 2-D multiweight multilength optical orthogonal code (MWML-OOC). Numerical results show that the Poisson approach underestimates the throughput performance in unacceptable levels and incorrectly predicts the number of successfully received packets for most offered load values even in favorable conditions, such as for the 2-D MWML-OOC OCDMA network with a considerably large number of simultaneous users. On the other hand, the binomial approach proved to be more straightforward, computationally more efficient, and just as accurate as the Markov chain approach.