In recent years, great progress has been made in the field of forecasting meteorological variables. Recently, deep learning architectures have made a major breakthrough in forecasting the daily average temperature over a ten-day horizon. However, advances in forecasting events related to the maximum temperature over short horizons remain a challenge for the community. A problem that is even more complex consists in making predictions of the maximum daily temperatures in the short, medium, and long term. In this work, we focus on forecasting events related to the maximum daily temperature over medium-term periods (90 days). Therefore, instead of addressing the problem from a meteorological point of view, this article tackles it from a climatological point of view. Due to the complexity of this problem, a common approach is to frame the study as a temporal classification problem with the classes: maximum temperature above normal, normal or below normal. From a practical point of view, we created a large historical dataset (from 1981 to 2018) collecting information from weather stations located in South America. In addition, we also integrated exogenous information from the Pacific, Atlantic, and Indian Ocean basins. We applied the AutoGluonTS platform to solve the above-mentioned problem. This AutoML tool shows competitive forecasting performance with respect to large operational platforms dedicated to tackling this climatological problem; but with a "relatively" low computational cost in terms of time and resources.

The efficiency of Deep Learning (DL) training jobs is critically dependent on dataloaders, which facilitate the transfer of data from storage to DL-accelerated hardware during training. Recent advancements in data loading technology have demonstrated significant improvements, not only in reducing training times but also in introducing capabilities such as seamless integration with cloud storage. This paper examines the dataloader as a distinct component within the DL workflow, offering a detailed analysis of its structure and functionalities. We present a systematic evaluation of various dataloading libraries, investigating their performance across different configurations, including worker count, batch size, GPU scaling, data access patterns and remote loading. The evaluation highlights trade-offs in functionality, usability, and performance. Additionally, we examine the impact of dataset characteristics on data loading performance, showing that throughput decreases exponentially with image resolution. To support ongoing research and practical advancements, we introduce the first open-source benchmarking suite for DL data loading, which allows the community to replicate, extend, and build upon our experiments.

We propose in this paper a coinvestment plan between several stakeholders of different types, namely a physical network owner, operating network nodes, e.g. a network operator or a tower company, and a set of service providers willing to use these resources to provide services as video streaming, augmented reality, autonomous driving assistance, etc. One such scenario is that of deployment of Edge Computing resources. Indeed, although the latter technology is ready, the high Capital Expenditure (CAPEX) cost of such resources is the barrier to its deployment. For this reason, a solid economical framework to guide the investment and the returns of the stakeholders is key to solve this issue. We formalize the coinvestment framework using coalitional game theory. We provide a solution to calculate how to divide the profits and costs among the stakeholders, taking into account their characteristics: traffic load, revenues, utility function. We prove that it is always possible to form the grand coalition composed of all the stakeholders, by showing that our game is convex. We derive the payoff of the stakeholders using the Shapley value concept, and elaborate on some properties of our game. We show our solution in simulation.

In this article, we introduce a fog system design for processing data collected from edge devices, such as mobile, sensor, and extended (mixed, augmented, virtual) reality equipment. Our system enables the network to provide hardware-accelerated processors for resource-intensive computations on data gathered from remote locations, such as 5G and beyond mobile networks. By splitting heavy computations into pipelines, and distributing them among processors in the edge, fog and the cloud, our design benefits from the processing power of the cloud, while utilizing fog devices with a lower network latency. We implement our design, and use it for distributed training and inference with industry-grade deep learning models for computer vision. We deploy our architecture in infrastructure including cloud and edge servers supporting GPU-accelerated computations. We benchmark pipelines in various deployment settings to study the overhead that they introduce. Our contributions are a new design for wide-area data processing, a framework that realizes this design and provides means of developing applications that are optimized in terms of infrastructure and hardware. These contributions are complemented with our benchmark results, which reveal the potential causes of processing overhead.

360-degree panoramic videos have gained considerable attention in recent years due to the rapid development of head-mounted displays (HMDs) and panoramic cameras. One major problem in streaming panoramic videos is that panoramic videos are much larger in size compared to traditional ones. Moreover, the user devices are often in a wireless environment, with limited battery, computation power, and bandwidth. To reduce resource consumption, researchers have proposed ways to predict the users’ viewports so that only part of the entire video needs to be transmitted from the server. However, the robustness of such prediction approaches has been overlooked in the literature: it is usually assumed that only a few models, pre-trained on past users’ experiences, are applied for prediction to all users. We observe that those pre-trained models can perform poorly for some users because they might have drastically different behaviors from the majority, and the pre-trained models cannot capture the features in unseen videos. In this work, we propose a novel meta learning based viewport prediction paradigm to alleviate the worst prediction performance and ensure the robustness of viewport prediction. This paradigm uses two machine learning models, where the first model predicts the viewing direction, and the second model predicts the minimum video prefetch size that can include the actual viewport. We first train two meta models so that they are sensitive to new training data, and then quickly adapt them to users while they are watching the videos. Evaluation results reveal that the meta models can adapt quickly to each user, and can significantly increase the prediction accuracy, especially for the worst-performing predictions.

In this paper, we study the impact of data complexity and data quality in the overlapping community detection problem. We show that community detection algorithms are very unstable against incomplete or erroneous data, and this result is consistent with all the evaluated performance metrics. We verify it using three quality metrics (F1, NMI, and Omega) when the ground-truth community structure is known, in four very popular and representative detection algorithms: Order Statistics Local Optimization Method (OSLOM), Greedy Clique Expansion (GCE) algorithm, Speaker-listener Label Propagation Algorithm (SLPA), and Cluster Affiliation Model for Big Networks (BIG-CLAM). We evaluate it over a set of real instances that arise from detecting the courses that belong to different careers (degrees) of an engineering University, and over large benchmark sets of synthetic instances frequently used in the literature.