Abstract
AOB is an Adaptive Optics (AO) facility currently designed to feed the Gemini infrared Multi Object Spectrograph (GIRMOS) on the GEMINI North 8m class telescope located in Hawaii. This AO system will be made of two AO modes. A laser tomography AO (LTAO) mode using 4 LGS (laser guide stars) and [1-3] NGS (natural guide stars) for high performance over a narrow field of view (a few a rcsec). The LTAO reconstruction will benefit from the most recent developments in the field, such as the super-resolution concept for the multi-LGS tomographic system, the calibration and optimization of the system on the sky, etc. The system will also operate in Ground Layer Adaptive Optics (GLAO) mode providing a robust solution for homogeneous partial AO correction over a wide 2' FOV. This last mode will also be used as a first s tep of a MOAO (Multi-object adaptive optics) mode integrated in the GIRMOS instrument. Both GLAO and LTAO modes are optimized to provide the best possible sky coverage, up to 60% at the North Galactic Pole. Finally, the project has been designed from day one as a fast-track, cost effective project, aiming to provide a first scientific light on the telescope by 2028 at the latest, with a good balance of innovative and creative concepts combined with standard and well controlled components and solutions. In this paper, we will present the innovative concepts, design and performance analysis of the two AO modes (LTAO and GLAO) of the AOB project.

Abstract
Chest X-Ray (CXR), plays a vital role in diagnosing lung and heart conditions, but the high demand for CXR examinations poses challenges for radiologists. Automatic support systems can ease this burden by assisting radiologists in the image analysis process. While Deep Learning models have shown promise in this task, concerns persist regarding their complexity and decision-making opacity. To address this, various visual explanation techniques have been developed to elucidate the model reasoning, some of which have received significant attention in literature and are widely used such as GradCAM. However, it is unclear how different explanations methods perform and how to quantitatively measure their performance, as well as how that performance may be dependent on the model architecture used and the dataset characteristics. In this work, two widely used deep classification networks - DenseNet121 and ResNet50 - are trained for multi-pathology classification on CXR and visual explanations are then generated using GradCAM, GradCAM++, EigenGrad-CAM, Saliency maps, LRP and DeepLift. These explanations methods are then compared with radiologist annotations using previously proposed explainability evaluations metrics - intersection over union and hit rate. Furthermore, a novel method to convey visual explanations in the form of radiological written reports is proposed, allowing for a clinically-oriented explainability evaluation metric - zones score. It is shown that Grad-CAM++ and Saliency methods offer the most accurate explanations and that the effectiveness of visual explanations is found to vary based on the model and corresponding input size. Additionally, the explainability performance across different CXR datasets is evaluated, highlighting that the explanation quality depends on the dataset's characteristics and annotations.

Abstract
This article proposes the Artificial Intelligence Models Switching Mechanism (AIMSM), a novel approach to optimize system resource utilization by allowing systems to switch AI models during runtime in dynamic environments. Many real-world applications utilize multiple data sources and various AI models for different purposes. In many of those applications, every AI model doesn’t have to operate all the time. The AIMSM strategically allows the system to activate and deactivate these models, focusing on system resource optimization. The switching of each AI model can be based on any information, such as context or previous results. In the case study of an autonomous mobile robot performing computer vision tasks, the AIMSM helps the system to achieve a significant increment in performance, with a 50% average increase in frames per second (FPS) rate, for this specific case study, assuming that no erroneous switching occurred. Experimental results have demonstrated that the AIMSM can improve system resource utilization efficiency when properly implemented, optimize overall resource consumption, and enhance system performance. The AIMSM presented itself as a better alternative to permanently loading all the models simultaneously, improving the adaptability and functionality of the systems. It is expected that using the AIMSM will yield a performance improvement that is particularly relevant to systems with multiple AI models of a complex nature, where such models do not need to be all continuously executed or systems that will benefit from lower resource usage. Code is available at https://github.com/BrunoGeorgevich/AIMSM.

Abstract
We present an overview of CheckThat! Lab's 2024 Task 3, which focuses on detecting 23 persuasion techniques at the text-span level in online media. The task covers five languages, namely, Arabic, Bulgarian, English, Portuguese, and Slovene, and highly-debated topics in the media, e.g., the Isreali-Palestian conflict, the Russia-Ukraine war, climate change, COVID-19, abortion, etc. A total of 23 teams registered for the task, and two of them submitted system responses which were compared against a baseline and a task organizers' system, which used a state-of-the-art transformer-based architecture. We provide a description of the dataset and the overall task setup, including the evaluation methodology, and an overview of the participating systems. The datasets accompanied with the evaluation scripts are released to the research community, which we believe will foster research on persuasion technique detection and analysis of online media content in various fields and contexts.

Abstract
Today, autonomous agents, the Internet of Things, and smart devices produce more and more distributed data and use them to learn models for different purposes. One challenge is that learning from local data only may lead to suboptimal models. Thus, better models are expected if agents can exchange data, leading to approaches such as federated learning. However, these approaches assume that data have no value and, thus, is exchanged for free. A machine learning data market (MLDM), a framework based on multiagent systems with a market-based perspective on data exchange, was recently proposed. In an MLDM, each agent trains its model based on both local data and data bought from other agents. Although the empirical results are interesting, several challenges are still open, including data acquisition and data valuation. The MLDM is an illustrative example of how the value of data can and should be integrated into the design of distributed ML systems.

Abstract
This paper present an instance transformation procedure to modify known instances of the resource -constrained project scheduling problem to make them easier to solve by heuristic and/or exact solution algorithms. The procedure makes use of a set of transformation rules that aim at reducing the feasible search space without excluding at least one possible optimal solution. The procedure will be applied to a set of 11,183 instances and it will be shown by a set of experiments that these transformations lead to 110 improved lower bounds, 16 new and better schedules (found by three meta -heuristic procedures and a set of branch -and -bound procedures) and even 64 new optimal solutions which were never not found before.