For more than five decades, Computer Science has remained one of the fundamental enabling technologies behind the continuous transformation of modern society. Industrial manufacturing, healthcare, education, and scientific research increasingly depend on advanced computing technologies to sustain innovation and progress. Likewise, everyday activities such as online shopping, social interaction, travel planning, and digital entertainment have become deeply intertwined with computational systems.
One of the most significant consequences of this technological evolution has been the unprecedented growth in data generation. The emergence of the Big Data era has produced massive datasets containing valuable information about complex natural, social, and technological processes. However, data alone has limited value unless effective methods are available to extract knowledge, identify patterns, and build predictive models. Machine Learning (ML), the discipline devoted to this objective, has consequently become one of the most active and transformative areas of research and innovation.
The success of modern ML has been driven by the convergence of two factors: the availability of large-scale datasets and the computational resources required to process them. ML technologies now influence a broad spectrum of applications, ranging from personalized content recommendation systems to critical domains such as healthcare, education, and scientific discovery. As a result, ML has become a key driver of technological progress and an increasingly important workload for modern computing systems.
At the same time, sustaining this progress presents significant challenges. The slowing of Moore’s Law, growing energy-efficiency constraints, and the increasing computational demands of large-scale neural networks require a fundamental rethinking of computer architectures. Future advances will depend on close collaboration between machine learning researchers and computer architects to design computing platforms capable of delivering higher performance while maintaining acceptable power and cost efficiency.
Within this context, the objective of this project is to explore novel microarchitectural techniques that improve the performance and energy efficiency of ML workloads executed on general-purpose processors. By advancing our understanding of the interaction between machine learning applications and modern hardware, the project aims to contribute both to scientific knowledge and to the development of technologies with potential industrial impact, fostering technology transfer and innovation in line with national and international R&D priorities.