Thesis offers

A list of issues hard to compute would include:

1. The computation of the partition function is NP-hard, involving an  exponential sum of terms
2. The exact computation of the derivative of the log-likelihood is  also NP-hard, since it contains the derivative of the partition function

Therefore, in practice we have to approximate both the computation of
the probabilities and several components of the learning process
itself. These drawbacks have prevented RBMs to show their real
potential as truly probabilistic models.

Currently, we are working on trying to improve several of the unsolved
issues related to RBMs:

1. - Mechanisms to control the learning process: www.lsi.upc.edu/~eromero/Publications/Downloads/2018-tnnls-stopcritRBM.pdf
2. Better approximations of the derivative of the log-likelihood: http://www.lsi.upc.edu/~eromero/Publications/Downloads/2019-nn-weightedCD.pdf
3. Efficient approximation of the partition function (work in progress)

These works have opened new lines of research, some of which can be
the topic of a Master's Thesis. The scope and degree of depth of the
work can be adapted to the estimated times to complete the Thesis. For
further details, contact Enrique Romero (eromero@cs.upc.edu).

Quantum computers promise exponential improvements over conventional ones due to the extraordinary properties of qubits. However, a quantum computer faces many challenges relative to the movement of qubits which is completely different from the movement of classical data. This thesis delves into these challenges and proposes solutions to create scalable quantum computing systems and the algorithms that run within them, following the current European projects at N3Cat (www.n3cat.upc.edu) on scalable quantum computing.

The interested candidate will work in a group of several PhD students and in collaboration with Universitat Politècnica de València, working in ONE of the following areas:

• Study the use of AI to model quantum computers or to assist in the development of architectural methods (qubit mapping, qubit partitioning, qubit routing).
• Study the potential scaling and adaptation of the existing quantum computing architectures to specific quantum machine learning algorithms (QAOA, VQE, others).

This is precisely why companies such as Google have invested a lot of resources on exploring applications of GNN, some of which are as mediatic as Alpha Fold (recently solving the Protein-Folding problem), or used by Goolge Maps to make predictions of the expected traffic of a road. In this regard, the goal of this thesis is to continue investigating about the potential of GNNs when applied to the field of Computer Networks. Our goal is to provide GNN-based solutions that make computer networks run more autonomously and in a more efficient way.

 The goal of this thesis is to enable the application of Deep Reinforcement Learning techniques for optimization problems in very large and complex scenarios, in this case Data Center Networks. Inspired by the last related work from OpenAI (https://openai.com/blog/evolution-strategies/), we will explore time-efficient alternatives to train a Graph Neural Network based DRL agent on large optimization scenarios in a reasonable amount of time.

A more detailed description of the project can be found in

https://www.cs.upc.edu/~eromero/Downloads/Retina-TFM-Project-01.pdf

The project is proposed in collaboration with Javier Zarranz Ventura
(Institut Clínic d'Oftalmologia, ICOF, Hospital Clínic de Barcelona, and
Institut d'Investigacions Biomèdiques August Pi I Sunyer, IDIBAPS),
which would provide a large annotated database to develop the project. For further information, please contact Alfredo Vellido (avellido@cs.
upc.edu) or Enrique Romero (eromero@cs.upc.edu).

Quantum computing is an emerging field poised to revolutionize information processing. Unlike classical computers that use bits to represent information as 0 or 1, quantum computers employ qubits, which can exist in superposed and entangled states. This enables more efficient solutions to complex problems, such as large number factorization or quantum system simulation.

However, quantum computing still faces significant challenges, including algorithm optimization. Many quantum processes require finding optimal solutions for optimization problems, such as minimizing system energy or determining molecular structures. This is where Deep Learning comes into play.

The goal of this research project is to explore how Deep Learning (in particular Deep Reinforcement Learning) techniques can enhance optimization in quantum computing. By applying deep neural networks, we can learn intricate patterns in quantum data and develop more efficient algorithms for solving optimization problems. These advancements could accelerate practical applications in areas like quantum cryptography, material simulation, and industrial process optimization.

You'll join an interdisciplinary team, since the work will be carried in collaboration with the Quantum Information Group at IMDEA Networks in Madrid. They will offer the expertise in Quantum  Physics while we work on the Deep Learning part..

 Additional details can be found in these papers:

 https://arxiv.org/abs/2310.01113

https://arxiv.org/abs/1910.07421

Recent advancements in nanotechnology have enabled the development of means for sensing and wireless communications with unprecedented miniaturization and capabilities, to the point that they can be introduced into the gastrointestinal tract inside a pill or into the bloodstream in the form of passively flowing nanomachines. 

This opens the door to the idea of intra-body communication networks, this is, a swarm of nanosensors inside the human body that use communications to coordinate their actions to sense and localize specific events (lack of oxygen, biomarkers, etc). This can lead to the development of applications such as continuous monitoring of diabetes, detection and localization of cancer micro-tumors, or early detection of blood clots. These possibilities are currently investigated by our team at the N3Cat (www.n3cat.upc.edu).

In this context, we are looking for excellent and self-motivated individuals who are eager to work on developing AI schemes (based on graph neural networks or multi-agent RL) for the detection and localization of events inside of the human body. Data will be gathered with an in-house simulator that integrates mobility models (BloodVoyagerS) and communication models (TeraSim).

The application of Artificial Intelligence (AI) and Machine Learning (ML) to network security (AI4SEC) is paramount against cybercrime. While AI/ML is mainstream in domains such as computer vision and natural language processing, traditional AI/ML has produced below-par results in AI4SEC. Solutions do not properly generalize, are ineffective in real deployments, and are vulnerable to adversarial attacks. A fundamental limitation is the lack of AI/ML technology specific to network security. Due to their unique ability to learn and generalize over graph-structured information, graph-learning approaches, and in particular Graph Neural Networks (GNNs), have recently enabled groundbreaking applications in multiple fields where data are generally represented as graphs. Network security data are intrinsically relational, and initial research suggests that graph-structured representations and GNNs have the potential to become foundational to AI4SEC, in the way convolutional and recursive networks were to computer vision and natural language processing.
The goal of GRAPHS4SEC is to leverage graph data representations and modern GNN technology to conceive a new breed of robust GNN-based network security methods which could radically advance the AI4SEC practice. The objectives of GRAPHS4SEC are: (a) to investigate algorithmic methods that facilitate modeling and learning from graph-based network security data; (b) to compare the benefits and overheads of GNN-based AI4SEC to traditional AI/ML in terms of detection performance, generalization, scalability, and robustness against adversarial attacks; (c) to showcase the benefits and improvements of GRAPHS4SEC technology in four critical, real-world network security applications with significant impact for society, considering (in particular) the detection and early mitigation of phishing and fake/malicious websites, a threat among the most popular and society-wide harmful in today's Internet.

As the field of deep learning continues to evolve, researchers are exploring novel ways to process and extract knowledge from complex data structures. One promising avenue is topological deep learning (TDL), which leverages topological domains such as simplicial complexes, cell complexes, and hypergraphs. TDL provides a comprehensive framework for analyzing data with intricate relationships, transcending traditional graph-based approaches. More information about TLD can be found here:

[2206.00606] Topological Deep Learning: Going Beyond Graph Data (arxiv.org)

On the other hand, IGNNITION is a powerful TensorFlow-based framework designed for fast prototyping of Graph Neural Networks (GNNs). It caters to network engineers, enabling them to create custom GNN models without delving into low-level TensorFlow code. With IGNNITION, scientists and engineers in the networking field can develop GNNs tailored to their specific problems within a matter of hours. IGNNITION is a VERY popular framework and accounts of thousands of downloads per month:

IGNNITION - Fast prototyping of Graph Neural Networks

ignnition · PyPI

Our research project aims to extend the IGNNITION framework to incorporate topological deep learning capabilities. Here are our specific objectives:

Combinatorial Complexes Integration: Enhance IGNNITION to handle combinatorial complexes, bridging the gap between graphs, hypergraphs, and cell complexes.

Message-Passing Combinatorial Complex Neural Networks (CCNNs): Develop a class of CCNNs within IGNNITION, emphasizing attention-based architectures.

Evaluation and Benchmarking: Assess the performance of CCNNs on mesh shape analysis and graph learning tasks, comparing them to existing state-of-the-art models.

The syntactic structure of a sentence can be represented as a tree where vertices are words and arcs indicate syntactic dependencies between words (Kübler et al 2012). In languages, these structures are typically planar, namely no two arcs cross when drawn above the sentence (Ferrer-i-Cancho et al 2018). Another property of these structures is that the distance between syntactically related words is much smaller than expected in a shuffling the sentence (Ferrer-i-Cancho et al 2022), a fact that lead to the formulation of the syntactic dependency distance minimization principle (Ferrer-i-Cancho 2004).

Syntactic dependency parsing is the branch of computational linguistic concerned with the extraction of syntactic dependency structures from raw text (Kübler et al 2012). This research proposal is focused on unsupervised syntactic dependency parsing, namely methods to extract syntactic dependency structures from unlabelled data (Han et al 2020). Although these methods can be extremely complex, here we aim to start from simple by taking Yuret's (1998) proposal as starting point to build from it more complex methods. Yuret's (1998) unsupervised parser consists of three steps:
1) training, i.e. learning the statistics of word co-occurrences in texts
2) generating a weighted graph where all vertices are the words that have appeared during training and weights are the mutual information between word pairs.
3) parsing a new sentence and thus obtaining its syntactic dependency structure. This structure is the maximum spanning tree from the subgraph of the weighted graph in Step 2 that is induced by the words of the sentence.

In Yuret's proposal, the maximum spanning tree in Step 3 is restricted to be planar. Interestingly, it has been argued that planarity may be a side-effect of pressure to reduce the distance between syntactically related word rather than a direct pressure of sentence to be planar (Gómez-Rodríguez & Ferrer-i-Cancho 2017, Gómez-Rodriguez, Christiansen & Ferrer-i-Cancho 2022). Therefore, the main goal of this project is to explore alternative unsupervised parsing methods:
a) an unsupervised parser that does not impose planarity. That is, Step 3 consists of computing a maximum spanning tree using classic algorithms for calculating maximum spanning trees.
b) an unsupervised parser that does not impose planarity but incorporates syntactic dependency distance minimization and thus tends to produce planar syntactic dependency structures naturally.

The bulk of the project consists of implementing simple parsers of the sort of a-b) and evaluating their performance. The parsers will be applied to human languages and also to sequences that other species produce (great apes and cetaceans).

This project will be developed in the environment of the LQMC research group https://lqmc.upc.edu/.

REFERENCES

Ferrer-i-Cancho, R (2004). Euclidean distance between syntactically linked words. Physical Review E 70, 056135.
Gómez-Rodríguez, C. & Ferrer-i-Cancho, R. (2017). Scarcity of crossing dependencies: a direct outcome of a specific constraint? Physical Review E 96, 062304.
Gómez-Rodríguez, C., Christiansen, M. & Ferrer-i-Cancho, R. (2022). Memory limitations are hidden in grammar. Glottometrics 52, 39-64.
Ferrer-i-Cancho, R. Gómez-Rodríguez, R., Esteban, J.L. (2018). Are crossing dependencies really scarce? Physica A: Statistical Mechanics and its Applications 493, 311-329.
Ferrer-i-Cancho, R., Gómez-Rodríguez & Esteban, J.L. & Alemany-Puig, L. (2022). Optimality of syntactic dependency distances. Physical Review E 105, 014308.
Han, W., Jiang, Y., Tou Ng, H. & Tu, K. (2020). A survey of unsupervised dependency parsing. In Proceedings of the 28th International Conference on Computational Linguistics, pages 2522¿2533, Barcelona, Spain. International Committee on Computational Linguistics.
Kübler, S., McDonald, R. & Nivre, J. (2009). Dependency parsing. Synthesis Lectures on Human Language Technologies 1, 1-127.
Yuret, D. (1998). Discovery of linguistic relations using lexical attraction. PhD Thesis, MIT.

Off-line policy learning in RL (from a dataset) is possible with off-policy methods when there is sufficient data. However, the quality of the policy will depend on the quality of the data obtained, and this quality depends (in part) on the amount of data collected. One possibility to improve the data obtained is to apply the knowledge of the symmetry of the system and the environment to do data augmentation of the dataset. In this work we will study:

1. Design of Off-line RL algorithms that take advantage of knowledge of the symmetries of the environment and the system.
2. Combine these systems to make successful initialization of on-policy methods.
3. Try to extend the knowledge of symmetries not by applying data augmentation on the dataset, but by using invariant neural networks.

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.

The research plan involves acquiring and preprocessing a diverse dataset of MRI scans, encompassing various tumor types, sizes, and locations for automatic tumor segmentation. The core of the project lies in training and fine-tuning the developed model with AI using already segmented tumor dataset. Our approach will explore the integration of multi-modal MRI data, such as T1-weighted, T2-weighted, and FLAIR images, to capture complementary information and improve segmentation accuracy Then, we will generalize the model across other patients'cohort and imaging protocols. 

Finally, to bridge the gap between research and clinical practice, the project will include the development of a user-friendly clinical toolbox to integrate the automated tumor segmentation algorithm in real world clinical scenarios.

LABINQUIRY project aims to develop a didactic infrastructure to help secondary school and university teachers to implement inquiry-based learning processes called study and research paths (SRPs). The starting point of an SRP is an open question that students address in small groups with the help of the teacher, they have to search for data, information, new tools and knowledge, etc, study them and jointly elaborate a single answer of the whole class.

One difficulty posed by the SRPs in terms of management is the teacher's treatment of the weekly reports produced by the student groups with the partial results of the work done during the week. The teacher responsible for the SRP has to "process" these reports in limited time in order to prepare the next class session. In this processing, the teacher has to read the reports - about 10-12 per class -, compile the most important information and group them according to the type of questions raised by each team, the data collected, the interim results provided, and the sources of information used.

The research group responsible for the LABINQUIRY project is interested in achieving an automatic processing of the content of student reports. The aim of the TFM would be, from a set of 10-12 reports, to collect and organise the information contained in the reports to help the teacher organise the next task. Both the questions posed by the teacher in the SRP and the students' reports are natural language texts. The collection and organisation of the information can be done at different levels and from different points of view (some of which are already fixed but some of which are not), giving flexibility to the work to be done during the TFM. In principle, the techniques to be used would fall into the following areas: unsupervised learning (clustering) and natural language processing (word embeddings).

The expected timing is January-June 2024.

To carry out this project it is advisable to have a good mastery of Spanish and/or Catalan and the techniques mentioned above. The work will be remunerated.

Seeing that not all neural networks fit to all problems, and that relational data is present in a wide variety of aspects of our daily life, the main focus of this thesis in N3Cat (www.n3cat.upc.edu) and BNN-UPC (www.bnn.upc.edu) is to explore the possibilities of the Graph Neural Networks (GNNs), whose aim is to learn and model graph-structured relational data. We are looking for students willing to study the uses, architectures, and algorithms of GNNs. To this end, the candidate will work on ONE of the following areas:

• Uses: Applying GNNs in emerging application areas, including but not limited to (1) electroencephalogram (EEG) analysis for Alzheimer's disease detection, epilepsy classification, motor imagery; (2) acceleration of the compilation of quantum computing algorithms
• Architectures: Investigating ways to accelerate the processing of GNNs in multiple computing platforms (CPU, GPU, accelerators).
• Algorithms: Developing meta-learning data-driven models to estimate the accuracy of a GNN for a given graph, without training, through synthetic graph generation and correlation analysis.