Abstract
This paper presents SpecRF-Posture, a novel low-cost approach for accurate Human Posture Recognition (HPR) using Radio Frequency (RF) signals. SpecRF-Posture leverages S21 parameters within the WiFi-6E frequency range for classification. We obtain a dataset of S21 parameters for different postures by performing beamscanning through mechanical rotation of a horn transmitter aimed at a reflective surface that illuminates the space of interest. We determine the S21 parameters of the signals that are then reflected back from the space onto an omni-directional receiver. Thus for each posture we attain the S21 parameters of each possible illumination direction of the space. Experimental results demonstrate that SpecRF-Posture achieves an accuracy of 99.17% in posture classification, highlighting its effectiveness. Additionally, an RF dataset was acquired using a software package for automatic data acquisition within the WiFi-6E frequency range, and both the dataset and the software package have been made publicly available.

Abstract
In the domain of RIS-based indoor localization, our work introduces two distinct approaches to address real-world challenges. The first method is based on deep learning, employing a Long Short-Term Memory (LSTM) network. The second, a novel LSTM-PSO hybrid, strategically takes advantage of deep learning and optimization techniques. Our simulations encompass practical scenarios, including variations in RIS placement and the intricate dynamics of multipath effects, all in Non-Line-of-Sight conditions. Our methods can achieve very high reliability, obtaining centimeter-level accuracy for the 98th percentile (worst case) in a different set of conditions, including the presence of the multipath effect. Furthermore, our hybrid approach showcases remarkable resolution, achieving submillimeter-level accuracy in numerous scenarios.

Abstract
Object detection is a crucial task in autonomous driving, where domain shift between the training and the test set is one of the main reasons behind the poor performance of a detector when deployed. Some erroneous priors may be learned from the training set, therefore a model must be invariant to conditions that might promote such priors. To tackle this problem, we propose an adversarial learning framework consisting of an encoder, an object-detector, and a condition-classifier. The encoder is trained to deceive the condition-classifier and aid the object-detector as much as possible throughout the learning stage, in order to obtain highly discriminative features. Experiments showed that this framework is not very competitive regarding the trade-off between precision and recall, but it does improve the ability of the model to detect smaller objects and some object classes.

Abstract
Phishing attacks aims to steal sensitive information and, unfortunately, are becoming a common practice on the web. Email phishing is one of the most common types of attacks on the web and can have a big impact on individuals and enterprises. There is still a gap in prevention when it comes to detecting phishing emails, as new attacks are usually not detected. The goal of this work was to develop a model capable of identifying phishing emails based on machine learning approaches. The work was performed in collaboration with E-goi, a multi-channel marketing automation company. The data consisted of emails collected from the E-goi servers in the electronic mail format. The problem consisted of a classification problem with unbalanced classes, with the minority class corresponding to the phishing emails and having less than 1% of the total emails. Several models were evaluated after careful data selection and feature extraction based on the email content and the literature regarding these types of problems. Due to the imbalance present in the data, several sampling methods based on under-sampling techniques were tested to see their impact on the model's ability to detect phishing emails. The final model consisted of a neural network able to detect more than 80% of phishing emails without compromising the remaining emails sent by E-goi clients.

Abstract
Emotions encompass physiological systems that can be assessed through biosignals like electromyography and electrocardiography. Prior investigations in emotion recognition have primarily focused on general population samples, overlooking the specific context of theatre actors who possess exceptional abilities in conveying emotions to an audience, namely acting emotions. We conducted a study involving 11 professional actors to collect physiological data for acting emotions to investigate the correlation between biosignals and emotion expression. Our contribution is the DECEiVeR (DatasEt aCting Emotions Valence aRousal) dataset, a comprehensive collection of various physiological recordings meticulously curated to facilitate the recognition of a set of five emotions. Moreover, we conduct a preliminary analysis on modeling the recognition of acting emotions from raw, low- and mid-level temporal and spectral data and the reliability of physiological data across time. Our dataset aims to leverage a deeper understanding of the intricate interplay between biosignals and emotional expression. It provides valuable insights into acting emotion recognition and affective computing by exposing the degree to which biosignals capture emotions elicited from inner stimuli.

Abstract
Extensive computational research has been dedicated to detecting keys and modes in tonal Western music within the major and minor modes. Little research has been dedicated to other modes and musical expressions, such as folk or non-Western music. This paper tackles this limitation by comparing traditional template-based with unsupervised machine-learning methods for diatonic mode detection within folk music. Template-based methods are grounded in music theory and cognition and use predefined profiles from which we compare a musical piece. Unsupervised machine learning autonomously discovers patterns embedded in the data. As a case study, the authors apply the methods to a dataset of Irish folk music called The Session on four diatonic modes: Ionian, Dorian, Mixolydian, and Aeolian. Our evaluation assesses the performance of template-based and unsupervised methods, reaching an average accuracy of about 80%. We discuss the applicability of the methods, namely the potential of unsupervised learning to process unknown musical sources beyond modes with predefined templates.