Consulta ofertes d'altres estudis i especialitats
The future of communication is taking us beyond traditional terrestrial networks. In the next decade, 6G networks will connect more than just smartphones-they will support flying cars and underwater robots using such high-tech enablers as lasers, satellites, drones, and high-altitude platforms (HAPs, flying at 20-40 km). To make this vision a reality, we need to tackle challenges like ensuring seamless connectivity for high-speed aerial vehicles, enabling underwater data transmission, and predicting large-scale network performance in urban environments.
In the NaNoNetworking Center in Catalunya (N3Cat, www.n3cat.upc.edu), we are investigating the use of ultra-short range wireless communications an alternative to current interconnects. In our EU project EWiC, we aim to emulate chip-to-chip wireless communications within a computer. For that, we will emulate the behavior of a multi-chip CPU with FPGAs, and connect them wirelessly using Software-Defined Radios (SDR).
Large language models (LLMs) are driven almost exclusively by industry corporations. Although model inference is typically accelerated on GPUs, there is an increasing demand for local inference, especially at the edge and fog computing, which instead uses CPUs equipped with recent hardware extensions such as VNNI and AMX to accelerate computations.
Recent advancements in nanotechnology have enabled the concept of the "Human Intranet", where devices inside and on our body can sense and communicate, opening the door to multiple exciting applications in the healthcare domain. This thesis aims to delve into the computing, communication, and localization aspects of the "Human Intranet" and how to practically realize them in the next decade.
State-of-the-art models such as LLMs are too large to fit in a single compute node (GPU, NPU, CPU), both for training and inference on a device (e.g., phone, laptop, tablet) or in larger-scale data centers. There is a need to develop optimization techniques to split and place these models onto a distributed set of compute nodes so that the overall system performance is maximized. The research will be focused on optimizing the placement of AI models onto distributed systems considering training time, energy consumption, and computational resources.
This project aims to improve the segmentation of bowel contents in abdominal CT images using MONAI Project. The proposed model will be evaluated and compared with segmentation results previously obtained using the nnU-Net framework, which were validated by medical experts for accuracy and clinical relevance.
A large-scale floating vehicle dataset of per-street segment traffic information, Metropolitan Segment Traffic Speeds from Massive Floating Car Data in 10 Cities (MeTS-10) is available for 10 global cities with a 15-minute resolution for collection periods ranging between 108 and 361 days in 2019¿2021 and covering more than 1500 square kilometers per metropolitan area. Data has been published by HERE. A comparison of the differences across some of the datasets in spatio-temporal coverage and variations in the reported traffic will be addressed in this master thesis.
This project focuses on the analysis and visualization of data obtained through manometry, a technique used to measure pressure within the small bowel. From the capture data, we aim to interpret the patterns of gastrointestinal motility. The project's goal is to enhance the understanding of various gastrointestinal disorders, through informative visual representations of the manometric data.
The High-Performance Linpack, also known as HPL, is a software package that solves a dense linear system and serves to assess the performance of distributed memory computers. Traditionally, it has used MPI and BLAS algorithms. Nowadays, it is ported to heterogeneous systems composed of accelerators such as GPUs.
This thesis aims to explore the possibilities of the new and less studied variant of neural networks called Graph Neural Networks (GNNs). While convolutional networks are good for computer vision or recurrent networks are good for temporal analysis, GNNs are able to learn and model graph-structured relational data, with huge implications in fields such as quantum chemistry, computer networks, or social networks among others.
Study existing techniques to reduce the computational footprint of Vision Transformers (ViT) while preserving accuracy. A set of efficiency improvement techniques will be selected and evaluated using a ViT-based test model to analyze their cost-accuracy trade-off.
Computing systems are ubiquitous in our daily life, to the point that progress is intimately tied to the improvements brought by new generations of the processors that lie at the heart of these systems. A common trait of current computing systems is that their internal data communication has become a fundamental bottleneck and traditional interconnects are just not good enough. This thesis aims to study how we can speed up architectures with CPUs, GPUs, and ML accelerators thanks to unconventional (e.g. wireless) interconnects.
Large language models (LLMs) are driven almost exclusively by industry corporations. The size of the model is currently a major limitation and quantisation arises as a technique to dramatically compress it, making it possible to perform training and inference at the edge and fog computing. This project consists of quantising LLMs to different low-bit precisions supported by modern CPU architectures while enhancing the hybrid parallelism of the executions. The ultimate goal is to improve the performance of these models.
The Digital Product Passport (DPP) will be an information system to do a better the circular economy of products. The information contained in the DPP of a product will help the consumers (and companies) to know many more details than today about a product, e.g., the origin of the materials, environmental information, repairability, how to recycle ... In Europe, the DPP will be an obligation for companies. A problem is that DPP systems in production do not yet exist.
This project aims to design novel ML-based advanced persistent threat (APT) detection algorithms. The student will work closely with industry experts from Telefónica on the development of tools and methods to combat APTs.
Development of a Hybrid Meta-heuristic to address the Dynamic Ride-Sharing Problem, combining Meta-heuristic optimization with Agent-based Simulation.
Bipolar Disorder (BD) is a psychiatric condition in which people experience significant shifts in mood, energy, and thought processes during manic and depressive episodes (Nierenberg et al., 2023). Manic episodes involve heightened energy, rapid speech, and grandiose thoughts, while depressive episodes are marked by low energy, slow speech, and feelings of hopelessness. The main objectives of this project are: i) To correlate speech features with manic-depressive symptoms severity in BD ii) To use speech features to develop predictive models for diagnostic (i.e., manic, de
Quantum computers promise exponential improvements over conventional ones due to the extraordinary properties of qubits. However, quantum computing faces many challenges relative to the scaling of the algorithms and of the computers that run them. This thesis delves into these challenges and proposes solutions to create scalable quantum computing systems.
The goal of this project is to improve crowd simulations by adding to it a realistic amount of animals, pets, and dogs in particular. While their presence might increase the overall realism and plausibility of the rendered simulations, and pedestrian trajectories might also be affected by it, urban designs could also benefit from the findings we might obtain.
This project focuses on extending an existing Virtual Reality (VR) simulation that illustrates the process of organ donation and transplantation. The main objective is to design and integrate an AI-powered assistant module into the current VR application. The AI assistant will act as a supportive, interactive guide throughout the simulation, offering real-time contextual information, answering participant questions, and adapting to user behavior to enhance engagement and understanding. will also evaluate
En base a dades recollides per un dispositiu IoT agrícola (humitat del sòl, consum d'aigua, variables climàtiques, etc.) l'objectiu és modelitzar i predir l'evolució de la humitat, detectar patrons crítics i anticipar les necessitats de reg, millorant així la planificació hídrica i optimitzant l'ús de l'aigua en entorns agrícoles.
This thesis aims to create a cost-effective, software-based eye-tracking system using high frequency sampling cameras. Such systems can democratize access to eye-tracking tools, which are typically expensive and require specialized hardware. The project will involve developing and validating algorithms capable of detecting and analyzing gaze direction, blinks, and fixations. Potential applications include educational research, human-computer interaction, and cognitive studies.
© Facultat d'Informàtica de Barcelona - Universitat Politècnica de Catalunya - Avís legal sobre aquest web - Configuració de privadesa