In recent years, Transformers have revolutionized the management of Natural Language Processing tasks, and Vision Transformers (ViTs) promise to do the same for Computer Vision ones. However, the adoption of ViTs is hampered by their computational cost. Indeed, given an image divided into patches, it is necessary to compute for each layer the attention of each patch with respect to all the others. Researchers have proposed many solutions to reduce the computational cost of attention layers by adopting techniques such as quantization, knowledge distillation and manipulation of input images. In this paper, we aim to contribute to the solution of this problem. In particular, we propose a new framework, called AgentViT, which uses Reinforcement Learning to train an agent that selects the most important patches to improve the learning of a ViT. The goal of AgentViT is to reduce the number of patches processed by a ViT, and thus its computational load, while still maintaining competitive performance. We tested AgentViT on CIFAR10, FashionMNIST, and Imagenette $$^+$$ + (which is a subset of ImageNet) in the image classification task and obtained promising performance when compared to baseline ViTs and other related approaches available in the literature.

The enormous growth of artificial intelligence (AI), and deep learning (DL) in particular, has led to the widespread use of these systems in a variety of contexts. One DL model capable of addressing complex computer vision tasks is the vision transformer (ViT). Despite its huge success, the reasoning behind the inferences it makes is often unclear, which poses significant challenges in critical scenarios. In this paper, we propose a new approach called MUltiplex Transformer EXplainer (MUTEX), which aims to explain the inferences made by ViTs. MUTEX combines multiplex network-based representations of attention matrices and mask perturbation approaches to provide insight into the inference process of ViTs. By mapping the attention layers of a ViT into a multiplex network, MUTEX is able to analyze the relationships between different parts of the input image and identify the image patches that most influence the inference process. We tested MUTEX on a subset of ImageNet and on BloodMNIST and compared its performance with that of existing visual explainability approaches. In addition, to assess the robustness and adaptability of MUTEX, we conducted a qualitative analysis, along with a hyperparameter and ablation study, which allowed us to further appreciate its potential in visual explainability of ViT.

Understanding how learners engage with immersive sign language training environments is essential for advancing virtual reality-based education and inclusion. This study analyzes behavioral engagement in SONAR, a virtual reality application designed for sign language training and validation. We focus on three automatically derived engagement indicators (Visual Attention (VA), Video Replay Frequency (VRF), and Post-Playback Viewing Time (PPVT)) and examine their relationship with learning performance in a sample of 117 university students. Participants completed a self-paced Training phase with 12 sign language instructional videos, followed by a Validation quiz assessing retention. We employed Pearson correlation analysis to examine the relationships between engagement indicators and quiz performance, followed by binomial Generalized Linear Model (GLM) regression to assess their joint predictive contributions. Additionally, we conducted temporal analysis by aggregating moment-to-moment VA traces across all learners to characterize engagement dynamics during the learning session. Results show that VA exhibits a strong positive correlation with quiz performance (r = 0.76), followed by PPVT (r = 0.66), whereas VRF shows no meaningful association. A binomial GLM confirms that VA and PPVT are significant predictors of learning success, jointly explaining a substantial proportion of performance variance ($pseudo-R^2$ = 0.83). Going beyond outcome-oriented analysis, we characterize temporal engagement patterns by aggregating moment-to-moment VA traces across all learners. The temporal profile reveals distinct attention peaks aligned with informationally dense segments of both training and validation videos, as well as phase-specific engagement dynamics, including initial acclimatization, oscillatory attention cycles during learning, and pronounced attentional peaks during assessment. Together, these findings highlight the central role of sustained and strategically allocated visual attention in VR-based sign language learning and demonstrate the value of behavioral trace data for understanding and predicting learner engagement in immersive environments.

Students are increasingly using Large Language Models (LLMs) to answer multiple-choice questions (MCQs) without engaging with the underlying concepts, which undermines the validity of traditional assessments. Instead of trying to detect or restrict LLM usage, an alternative strategy is redesigning MCQs so that relying on LLMs becomes ineffective. In this paper, we propose GradQuiz, a framework that generates adversarial distractors targeting the decision-making process of LLMs. GradQuiz leverages gradient-based signals from a target LLM to perturb semantically influential entities in an MCQ. This process generates plausible distractors, which are then refined to ensure grammatical and semantic coherence. We evaluated GradQuiz on two MCQ benchmarks, i.e., OpenTriviaQA and Massive Multitask Language Understanding (MMLU), across multiple LLM families, including both open-weight and proprietary black-box models. Results show that GradQuiz consistently reduces LLM accuracy in answering MCQs. In particular, when applied to gemma-3-27b-it, it reduces accuracy by 67.26% compared to the original quizzes and by 51.07% compared to the strongest competing adversarial distractor approach on the MMLU dataset, while it achieves accuracy reductions of 49.31% and 42.92%, respectively, on the OpenTriviaQA dataset. Human evaluations confirm that the generated distractors preserve pedagogical coherence and relevance, while a controlled study on students shows that GradQuiz does not increase MCQ difficulty for learners who do not rely on LLMs to answer quizzes.