Prior work on scientific question answering has largely emphasized chatbot-style systems, with limited exploration of fine-tuning foundation models for domain-specific reasoning. In this study, we developed a chatbot for the University of Limerick's Department of Electronic and Computer Engineering to provide course information to students. A custom dataset of 1,203 question-answer pairs in SQuAD format was constructed using the university book of modules, supplemented with manually and synthetically generated entries. We fine-tuned BERT (Devlin et al., 2019) using PyTorch and evaluated performance with Exact Match and F1 scores. Results show that even modest fine-tuning improves hypothesis framing and knowledge extraction, demonstrating the feasibility of adapting foundation models to educational domains. While domain-specific BERT variants such as BioBERT and SciBERT exist for biomedical and scientific literature, no foundation model has yet been tailored to university course materials. Our work addresses this gap by showing that fine-tuning BERT with academic QA pairs yields effective results, highlighting the potential to scale towards the first domain-specific QA model for universities and enabling autonomous educational knowledge systems.

This is the focus of work by and, which won the best paper award at the recent RoboCup symposium . The symposium takes place alongside the annual RoboCup competition, which this year was held in Salvador, Brazil. We caught up with some of the authors to find out more about the work, how their method can be transferred to applications beyond RoboCup, and their future plans for the competition. Could you start by giving us a brief description of the problem that you were trying to solve in your paper "Self-supervised Feature Extraction for Enhanced Ball Detection on Soccer Robots"? The main challenge we faced was that deep learning generally requires a large amount of labeled data. This is not a major problem for common tasks that have already been studied, because you can usually find labeled datasets online.

This paper presents Project Riley, a novel multimodal and multi-model conversational AI architecture oriented towards the simulation of reasoning influenced by emotional states. Drawing inspiration from Pixar's Inside Out, the system comprises five distinct emotional agents - Joy, Sadness, Fear, Anger, and Disgust - that engage in structured multi-round dialogues to generate, criticise, and iteratively refine responses. A final reasoning mechanism synthesises the contributions of these agents into a coherent output that either reflects the dominant emotion or integrates multiple perspectives. The architecture incorporates both textual and visual large language models (LLMs), alongside advanced reasoning and self-refinement processes. A functional prototype was deployed locally in an offline environment, optimised for emotional expressiveness and computational efficiency. From this initial prototype, another one emerged, called Armando, which was developed for use in emergency contexts, delivering emotionally calibrated and factually accurate information through the integration of Retrieval-Augmented Generation (RAG) and cumulative context tracking. The Project Riley prototype was evaluated through user testing, in which participants interacted with the chatbot and completed a structured questionnaire assessing three dimensions: Emotional Appropriateness, Clarity and Utility, and Naturalness and Human-likeness. The results indicate strong performance in structured scenarios, particularly with respect to emotional alignment and communicative clarity.

Malware evolves rapidly, forcing machine learning (ML)-based detectors to adapt continuously. With antivirus vendors processing hundreds of thousands of new samples daily, datasets can grow to billions of examples, making full retraining impractical. Continual learning (CL) has emerged as a scalable alternative, enabling incremental updates without full data access while mitigating catastrophic forgetting. In this work, we analyze a critical yet overlooked issue in this context: security regression. Unlike forgetting, which manifests as a general performance drop on previously seen data, security regression captures harmful prediction changes at the sample level, such as a malware sample that was once correctly detected but evades detection after a model update. Although often overlooked, regressions pose serious risks in security-critical applications, as the silent reintroduction of previously detected threats in the system may undermine users' trust in the whole updating process. To address this issue, we formalize and quantify security regression in CL-based malware detectors and propose a regression-aware penalty to mitigate it. Specifically, we adapt Positive Congruent Training (PCT) to the CL setting, preserving prior predictive behavior in a model-agnostic manner. Experiments on the ELSA, Tesseract, and AZ-Class datasets show that our method effectively reduces regression across different CL scenarios while maintaining strong detection performance over time.

We propose a ``half'' layer of hidden units that has some of its weights randomly set and some of them trained. A half unit is composed of two stages: First, it takes a weighted sum of its inputs with fixed random weights, and second, the total activation is multiplied and then translated using two modifiable weights, before the result is passed through a nonlinearity. The number of modifiable weights of each hidden unit is thus two and does not depend on the fan-in. We show how such half units can be used in the first or any later layer in a deep network, possibly following convolutional layers. Our experiments on MNIST and FashionMNIST data sets indicate the promise of half layers, where we can achieve reasonable accuracy with a reduced number of parameters due to the regularizing effect of the randomized connections.

The implications of such models can apply to real-world computers, as long as resource utilization does not exceed their physical limitations. Even when those bounds are reached, there is still the question of what could in the future be computed on machines of ever-greater size and speed (https://bit.ly/3FiNjgW). However, when even futuristic physical limitations and issues like power consumption are addressed, the correspondence between the infinitary models and reality starts to fray. A widely understood example of this divergence can be found in the application of the theory of algorithmic complexity to sorting. The classical analysis of sorting yields the well-known result (https://bit.ly/3D7gIKE)

Accurately estimating the position of a patient's side robotic arm in real time during remote surgery is a significant challenge, especially within Tactile Internet (TI) environments. This paper presents a new and efficient method for position estimation using a Kalman Filter (KF) combined with the Multivariable Output-Error State Space (MOESP) method for system identification. Unlike traditional approaches that require prior knowledge of the system's dynamics, this study uses the JIGSAW dataset, a comprehensive collection of robotic surgical data, along with input from the Master Tool Manipulator (MTM) to derive the state-space model directly. The MOESP method allows accurate modeling of the Patient Side Manipulator (PSM) dynamics without prior system models, improving the KF's performance under simulated network conditions, including delays, jitter, and packet loss. These conditions mimic real-world challenges in Tactile Internet applications. The findings demonstrate the KF's improved resilience and accuracy in state estimation, achieving over 95 percent accuracy despite network-induced uncertainties.

As wireless researchers are tasked to enable wireless communication as infrastructure in more dynamic aerial settings, there is a growing need for large-scale experimental platforms that provide realistic, reproducible, and reliable experimental validation. To bridge the research-to-implementation gap, the Aerial Experimentation and Research Platform for Advanced Wireless (AERPAW) offers open-source tools, reference experiments, and hardware to facilitate and evaluate the development of wireless research in controlled digital twin environments and live testbed flights. The inaugural AERPAW Challenge, "Find a Rover," was issued to spark collaborative efforts and test the platform's capabilities. The task involved localizing a narrowband wireless signal, with teams given ten minutes to find the "rover" within a twenty-acre area. By engaging in this exercise, researchers can validate the platform's value as a tool for innovation in wireless communications research within aerial robotics. This paper recounts the methods and experiences of the top three teams in automating and rapidly locating a wireless signal by automating and controlling an aerial drone in a realistic testbed scenario.

Glaucoma is a chronic eye disease characterized by optic neuropathy, leading to irreversible vision loss. It progresses gradually, often remaining undiagnosed until advanced stages. Early detection is crucial to monitor atrophy and develop treatment strategies to prevent further vision impairment. Data-centric methods have enabled computer-aided algorithms for precise glaucoma diagnosis. In this study, we use deep learning models to identify complex disease traits and progression criteria, detecting subtle changes indicative of glaucoma. We explore the structure-function relationship in glaucoma progression and predict functional impairment from structural eye deterioration. We analyze statistical and machine learning methods, including deep learning techniques with optical coherence tomography (OCT) scans for accurate progression prediction. Addressing challenges like age variability, data imbalances, and noisy labels, we develop novel semi-supervised time-series algorithms: 1. Weakly-Supervised Time-Series Learning: We create a CNN-LSTM model to encode spatiotemporal features from OCT scans. This approach uses age-related progression and positive-unlabeled data to establish robust pseudo-progression criteria, bypassing gold-standard labels. 2. Semi-Supervised Time-Series Learning: Using labels from Guided Progression Analysis (GPA) in a contrastive learning scheme, the CNN-LSTM architecture learns from potentially mislabeled data to improve prediction accuracy. Our methods outperform conventional and state-of-the-art techniques.

In the competitive realm of sports, optimal performance necessitates rigorous management of nutrition and physical conditioning. Specifically, in badminton, the agility and precision required make it an ideal candidate for motion analysis through video analytics. This study leverages advanced neural network methodologies to dissect video footage of badminton matches, aiming to extract detailed insights into player kinetics and biomechanics. Through the analysis of stroke mechanics, including hand-hip coordination, leg positioning, and the execution angles of strokes, the research aims to derive predictive models that can suggest improvements in stance, technique, and muscle orientation. These recommendations are designed to mitigate erroneous techniques, reduce the risk of joint fatigue, and enhance overall performance. Utilizing a vast array of data available online, this research correlates players' physical attributes with their in-game movements to identify muscle activation patterns during play. The goal is to offer personalized training and nutrition strategies that align with the specific biomechanical demands of badminton, thereby facilitating targeted performance enhancements.