Rubén Cuevas

Francisco Caravaca, Ángel Cuevas, Rubén Cuevas

The rapid expansion of Large Language Models (LLMs) has introduced unprecedented energy demands, extending beyond training to large-scale inference workloads that often dominate total lifecycle consumption. Deploying these models requires energy-intensive GPU infrastructure, and in some cases has even prompted plans to power data centers with nuclear energy. Despite this growing relevance, systematic analyses of inference energy consumption remain limited. In this work, we present a large-scale measurement-based study comprising over 32,500 measurements across 21 GPU configurations and 155 model architectures, from small open-source models to frontier systems. Using the vLLM inference engine, we quantify energy usage at the prompt level and identify how architectural and operational factors shape energy demand. Building on these insights, we develop a predictive model that accurately estimates inference energy consumption across unseen architectures and hardware, and implement it as a browser extension to raise awareness of the environmental impact of generative AI.

José González Cabañas, Patricia Callejo, Rubén Cuevas et al.

Energy is today the most critical environmental challenge. The amount of carbon emissions contributing to climate change is significantly influenced by both the production and consumption of energy. Measuring and reducing the energy consumption of services is a crucial step toward reducing adverse environmental effects caused by carbon emissions. Millions of websites rely on online advertisements to generate revenue, with most websites earning most or all of their revenues from ads. As a result, hundreds of billions of online ads are delivered daily to internet users to be rendered in their browsers. Both the delivery and rendering of each ad consume energy. This study investigates how much energy online ads use in the rendering process and offers a way for predicting it as part of rendering the ad. To the best of the authors' knowledge, this is the first study to calculate the energy usage of single advertisements in the rendering process. Our research further introduces different levels of consumption by which online ads can be classified based on energy efficiency. This classification will allow advertisers to add energy efficiency metrics and optimize campaigns towards consuming less possible.

Jorge Marco-Blanco, Rubén Cuevas

Time-series classification is essential across diverse domains, including medical diagnosis, industrial monitoring, financial forecasting, and human activity recognition. The Rocket algorithm has emerged as a simple yet powerful method, achieving state-of-the-art performance through random convolutional kernels applied to time-series data, followed by non-linear transformation. Its architecture approximates a one-hidden-layer convolutional neural network while eliminating parameter training, ensuring computational efficiency. Despite its empirical success, fundamental questions about its theoretical foundations remain unexplored. We bridge theory and practice by formalizing Rocket's random convolutional filters within the compressed sensing framework, proving that random projections preserve discriminative patterns in time-series data. This analysis reveals relationships between kernel parameters and input signal characteristics, enabling more principled approaches to algorithm configuration. Moreover, we demonstrate that its non-linearity, based on the proportion of positive values after convolutions, expresses the inherent sparsity of time-series data. Our theoretical investigation also proves that Rocket satisfies two critical conditions: translation invariance and noise robustness. These findings enhance interpretability and provide guidance for parameter optimization in extreme cases, advancing both theoretical understanding and practical application of time-series classification.

Jorge Marco-Blanco, Rubén Cuevas

Time series data, spanning applications ranging from climatology to finance to healthcare, presents significant challenges in data mining due to its size and complexity. One open issue lies in time series clustering, which is crucial for processing large volumes of unlabeled time series data and unlocking valuable insights. Traditional and modern analysis methods, however, often struggle with these complexities. To address these limitations, we introduce R-Clustering, a novel method that utilizes convolutional architectures with randomly selected parameters. Through extensive evaluations, R-Clustering demonstrates superior performance over existing methods in terms of clustering accuracy, computational efficiency and scalability. Empirical results obtained using the UCR archive demonstrate the effectiveness of our approach across diverse time series datasets. The findings highlight the significance of R-Clustering in various domains and applications, contributing to the advancement of time series data mining.

Antonio Pastor, Rubén Cuevas, Ángel Cuevas et al.

Programmatic advertising operates one of the most sophisticated and efficient service platforms on the Internet. However, the complexity of this ecosystem is a direct cause of one of the most important problems in online advertising, the lack of transparency. This lack of transparency enables subsequent problems such as advertising fraud, which causes billions of dollars in losses. In this paper we propose Ads.chain, a technological solution to the lack-of-transparency problem in programmatic advertising. Ads.chain extends the current effort of the Internet Advertising Bureau (IAB) in providing traceability in online advertising through the Ads.txt and Ads.cert solutions, addressing the limitations of these techniques. Ads.chain is (to the best of the authors' knowledge) the first solution that provides end-to-end cryptographic traceability at the ad transaction level. It is a communication protocol that can be seamlessly embedded into ad-tags and the OpenRTB protocol, the de-facto standards for communications in online advertising, allowing an incremental adoption by the industry. We have implemented Ads.chain and made the code publicly available. We assess the performance of Ads.chain through a thorough analysis in a lab environment that emulates a real ad delivery process at real-life throughputs. The obtained results show that Ads.chain can be implemented with limited impact on the hardware resources and marginal delay increments at the publishers lower than 0.20 milliseconds per ad space on webpages and 2.6 milliseconds at the programmatic advertising platforms. These results confirm that Ads.chain's impact on the user experience and the overall operation of the programmatic ad delivery process can be considered negligible.

Aritz Arrate, José González Cabañas, Ángel Cuevas et al.

Online advertising is the major source of income for a large portion of Internet Services. There exists a body of literature aiming at optimizing ads engagement, understanding the privacy and ethical implications of online advertising, etc. However, to the best of our knowledge, no previous work analyses at large scale the exposure of real users to online advertising. This paper performs a comprehensive analysis of the exposure of users to ads and advertisers using a dataset including more than 7M ads from 140K unique advertisers delivered to more than 5K users that was collected between October 2016 and May 2018. The study focuses on Facebook, which is the second largest advertising platform only to Google in terms of revenue, and accounts for more than 2.2B monthly active users. Our analysis reveals that Facebook users are exposed (in median) to 70 ads per week, which come from 12 advertisers. Ads represent between 10% and 15% of all the information received in users' newsfeed. A small increment of 1% in the portion of ads in the newsfeed could roughly represent a revenue increase of 8.17M USD per week for Facebook. Finally, we also reveal that Facebook users are overprofiled since in the best case only 22.76% of the interests Facebook assigns to users for advertising purpose are actually related to the ads those users receive.