Ofertes de projectes

Consulta ofertes d'altres estudis i especialitats

Boltzmann Machines are probabilistic models developed in 1985 by D.H. Ackley, G.E. Hinton and T.J. Sejnowski. In 2006, Restricted Boltzmann Machines (RBMs) were used in the pre-training step of several successful deep learning models, leading to a new renaissance of neural networks and artificial intelligence. In spite of their nice mathematical formulation, there are a number of issues that are hard to compute. This project aims to address any of these issues.

In this project, we wish to create a summarized universal representation of packet flows within a computer network. We approach this problem as an Unsupervised Learning problem, where the summarized representation must be as small as possible while minimizing the reconstruction error. In order to build this representation, multiple approaches are to be considered. This includes using traditional techniques such as a Fourier Transformation, and using representations learned Machine Learning models, specifically Autoencoders and sequence models like 1-dimensional CNNs, RNNs

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.

To date, traditional Deep Learning (DL) solutions (e.g. Feed-forward Neural Networks, Convolutional Neural Networks) have had a major impact in numerous fields, such as Speak Recognition (e.g., Siri, Alexa), Autonomous driving, Computer Vision,etc. It was just recently, however, that a new DL technique called Graph Neural Network (GNN) was introduced, proving to be unprecedentedly accurate to solve problems that are formalized as graphs.

Recent advances in the field of Reinforcement Learning (DRL) are rising a lot of attention due to its potential for automatic control and automatization. Breakthroughs from academia and the industry (e.g, Stanford, DeepMind and OpenAI) are demonstrating that DRL is an effective technique to face complex optimization problems with many dimensions and non-linearities. However, to train a DRL agent in large optimization scenarios still remains a challenge due to the computational intensive operations during backpropagation.

This project aims to analyze the prediction capability of Optical Coherence Tomography Angiography (OCTA) images for Diabetes Mellitus (DM) and Diabetic Retinopathy (DR,) in a large high-quality image dataset from previous research projects carried out in the field of Ophthalmology (Fundacio¿ La Marato¿ TV3, Fondo Investigaciones Sanitarias, FIS). OCTA is a newly developed, non-invasive, retinal imaging technique that permits adequate delineation of the perifoveal vascular network. It allows the detection of paramacular areas of capillary non perfusion and/or enlargement of the foveal avascular zone (FAZ), representing an excellent tool for assessment of DR.

Ciència de les Dades i Intel·ligència Computacional Visió, Percepció i Robòtica. Tecnologies Assistencials

LiDAR sensors, crucial in robotics and autonomous driving, emit laser beams to estimate the 3D position of reflected points, stored as a "point cloud". Deep Learning techniques are used in Computer Vision for tasks like image classification, object detection, and semantic segmentation. Recently, these techniques have been applied to point clouds instead of images, expanding applications, especially in autonomous navigation, due to LiDAR's precision and robustness. The goal of this Master's thesis is to investigate Deep Learning techniques for the Semantic Segmentation.

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.

S'ofereix una beca d'iniciació a la recerca de 20 hores/setmana amb un salari aprox. de 600 Euros/mes per realitzar el TFM en el marc del projecte GRAPHSEC.

The syntactic structure of a sentence can be represented as a tree where vertices are words and arcs indicate syntactic dependencies between words. Syntactic dependency parsing is the branch of computational linguistic concerned with the extraction of syntactic dependency structures from raw text. This research proposal is focused on unsupervised syntactic dependency parsing, i.e. methods to extract syntactic dependency structures from unlabelled data. This projects consists of implementing simple unsupervised parsers and evaluating them on human languages and other species

The goal of the project consists in the design of off-line RL algorithms that leverage knowledge about the symmetry of the environment and system, and to obtain better policies with fewer data.

Ciència de les Dades i Intel·ligència Computacional Enginyeria del Coneixement i Aprenentatge Automàtic

This research proposal addresses the critical need for efficient and accurate tumor segmentation in neuroimaging, particularly utilizing Magnetic Resonance Imaging (MRI) data. The project aims to develop a robust and automated approach for tumor delineation. The proposed methodology integrates state-of-the-art MRI neuroimaging techniques with advanced artificial learning methods.

This work is included in the research project LABINQUIRY, in a teaching environment. The goal is to develop a system capable of collecting and organise documents (pairs question-answer) expessed in natural language. These documents are generated within the interaction between the teacher and the students.

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.

Consulta ofertes d'altres estudis i especialitats