Human Activity Recognition (HAR) is an essential area of research related to the ability smartphones retrieve information through embedded sensors and recognize activity that humans are performing. Researchers have recognized people's activities by processing data received from with Machine Learning Models. This work intended be a hands-on survey practical's tables capable guiding reader used in modern highly cited developed machine learning models perform human recognition. Several papers...

As the occurrence of Denial Service and Distributed (DoS/DDoS) attacks increases, demand for effective defense mechanisms increases. Recognition such anomalies in computer network is commonly performed through network-based intrusion detection prevention systems (NIDPSs). Although NIDPSs allow interception all known attacks, they are not robust to continuing variation over time DoS/DDoS anomalies. The machine learning (ML) paradigm provides algorithms that can effectively reduce concept...

Automatic traffic flow classification is useful to reveal road congestions and accidents. Nowadays, roads highways are equipped with a huge amount of surveillance cameras, which can be used for real-time vehicle identification, thus providing estimation. This research provides comparative analysis state-of-the-art object detectors, visual features, models implement state estimations. More specifically, three different detectors compared identify vehicles. Four machine learning techniques...

The 0-day attack is a cyber-attack based on vulnerabilities that have not yet been published. detection of anomalous traffic generated by such attacks vital, as it can represent critical problem, both in technical and economic sense, for smart enterprise any system largely dependent technology. To predict this kind attack, one solution be to use unsupervised machine learning approaches, they guarantee the anomalies regardless their prior knowledge. It also essential identify unknown...

Recent enhancements in Large Language Models (LLMs) such as ChatGPT have exponentially increased user adoption. These models are accessible on mobile devices and support multimodal interactions, including conversations, code generation, patient image uploads, broadening their utility providing healthcare professionals with real-time for clinical decision-making. Nevertheless, many authors highlighted serious risks that may arise from the adoption of LLMs, principally related to safety...

Recognition of known malicious patterns through signature-based systems is unsuccessful against malware for which no signature exists to identify them. These include not only zero-day but also software able self-replicate rewriting its own code leaving unaffected execution, namely metamorphic malware. YARA a popular analysis tool that uses the so-called YARA-rules, are built match contents within files or network packets analyzed by an Anti-Virus engine. Sometimes such content expressed in...

This study used Explainable Artificial Intelligence (XAI) with SHapley Additive exPlanations (SHAP) to examine dietary and lifestyle habits in the Spanish population identify key diet predictors. A cross-sectional design was used, employing validated NutSo-HH scale gather data on nutrition, lifestyle, socio-demographic factors. The CatBoost method combined SHAP applied. sample included 22,181 adults: 17,573 followed Mediterranean diet, 1425 were vegetarians, 365 vegans, 1018 practiced...

The widespread use of artificial intelligence deep neural networks in fields such as medicine and engineering necessitates understanding their decision-making processes. Current explainability methods often produce inconsistent results struggle to highlight essential signals influencing model inferences. This paper introduces the Evolutionary Independent Deterministic Explanation (EVIDENCE) theory, a novel approach offering deterministic, model-independent method for extracting significant...

Neurodegenerative diseases are particular whose decline can partially or completely compromise the normal course of life a human being. In order to increase quality patient's life, timely diagnosis plays major role. The analysis neurodegenerative diseases, and their stage, is also carried out by means gait analysis. Performing early stage disease assessment still an open problem. this paper, focus on modeling movement pattern using kinematic theory rapid movements its sigma-lognormal model....

Recognition of malware is critical in cybersecurity as it allows for avoiding execution and the downloading malware. One possible approaches to analyze executable’s Application Programming Interface (API) calls, which can be done using tools that work sandboxes, such Cuckoo or CAPEv2. This chain calls then used classify if considered file benign aims compare six modern shallow learning deep techniques based on tabular data, two datasets API containing goodware, where corresponding expressed...

Packet classification activity performed by a FireWall (FW) introduces high latency in network communications due to the computation time required check whether any packet matches one of FW rules. Such process is done sequentially checking list rules until match found or end reached. Given complexity some environments, this could become relevant. This problem addressed ordering minimize latency, where with higher activation frequencies are placed accordingly starting from top list. not...

A web application is prone to security threats due its open nature. The of these platforms imperative for organizations all sizes because they store sensitive information. Consequently, exploiting vulnerabilities could result in large-scale data breaches and significant brand financial damages. SQL injection (SQLi) represents a popular attack vector that malicious actors use compromise website security. Web firewalls (WAFs) play primary role preventing such typologies. In the recent...

Vehicular traffic flow prediction for a specific day of the week in time span is valuable information. Local police can use this information to preventively control more critical areas and improve viability by decreasing, also, number accidents. In paper, novel generative deep learning architecture series analysis, inspired Google DeepMind’ Wavenet network, called TrafficWave, proposed applied problem. The technique compared with most performing state-of-the-art approaches: stacked auto...

In a society with increasing age, the understanding of human falls it is paramount importance. This paper presents Decision Support System whose pipeline designed to extract and compute physical domain's features achieving state art accuracy on Le2i UR fall detection datasets. The uses Kinematic Theory Rapid Human Movement its sigma-lognormal model together classic achieve 98% 99% in automatic respectively URFD effort made design this work toward recognition by using models laws are clear...

Deep learning (DL) has been demonstrated to be a valuable tool for analyzing signals such as sounds and images, thanks its capabilities of automatically extracting relevant patterns well end-to-end training properties. When applied tabular structured data, DL exhibited some performance limitations compared shallow techniques. This work presents novel technique data called adaptive multiscale attention deep neural network architecture (also named excited attention). By exploiting parallel...

Multiclass classification in cancer diagnostics, using DNA or Gene Expression Signatures, but also of bacteria species fingerprints MALDI-TOF mass spectrometry data, is challenging because imbalanced data and the high number dimensions with respect to instances. In this study, a new oversampling technique called LICIC will be presented as valuable instrument countering both class imbalance, famous “curse dimensionality” problem. The method enables preservation non-linearities within dataset,...

This benchmarking study aims to examine and discuss the current state-of-the-art techniques for in-video violence detection, also provide results as a reference future accuracy baseline of detection systems. In this paper, authors review 11 detection. They re-implement five carefully chosen over three different publicly available datasets, using several classifiers, all in same conditions. The main contribution work is compare feature-based modern deep-learning techniques, such Inception V3.