Research Project: the flower paradigm and transformer classifier for EEG data

Research Project: the flower paradigm and transformer classifier for EEG data

The flower paradigm and a transformer classifier for EEG dataThe primary objective of the project is the development of a transformer model for the classification of EEG-related data. Subsequently, our inquiry delves into the crucial aspect of preprocessing, seeking to elucidate its pivotal role in influencing the overall efficacy and performance of the aforementioned architecture. To rigorously examine this dimension, we intend to undertake a systematic comparative analysis. This entails the initial classification task on extensively preprocessed data, followed by a parallel evaluation on data subjected to minimal preprocessing, and ultimately, the assessment of the classifier's effectiveness on raw, unprocessed data. Looking beyond the immediate confines of our research focus, the broader vision of the work is oriented towards pioneering a fresh and innovative paradigm within the domain of assistive communication technology, particularly within the context of speller technology designed for individuals with communication impairments as ALS patients. This paradigm represents a pivotal shift, characterized by a fundamental reliance on the...
Read More
Research Project on Continual Learning

Research Project on Continual Learning

by Luca Bottero Our interests center around the intriguing domain of Continual Learning, a subset of machine learning focused on training models to adapt and evolve over time. This involves enabling these models to learn from new data while retaining previously acquired knowledge. In today’s fast-paced world, characterized by constant changes in data and knowledge, Continual Learning assumes primary importance. Traditional machine learning models often encounter challenges when confronted with novel information, a phenomenon commonly referred to as catastrophic forgetting. Continual Learning presents a remedy to this issue by facilitating continuous learning, rendering models adaptable, efficient, and adept at handling evolving data environments. Currently, we are working in a project that applies these Continual Learning techniques to gravitational wave interferometer data. We aspire to develop adaptive models that can learn from new experimental runs while retaining knowledge learned from prior detections. In a world characterized by relentless technological advancements and cease- less data generation, the concept of Continual Learning remains pivotal. The Turin...
Read More
Machine Learning for Anomaly detection in Prompt Gamma Spectra

Machine Learning for Anomaly detection in Prompt Gamma Spectra

by Beatrice Villata Tumor treatment can be achieved with radiotherapy techniques by irradiating the target volumes with ionizing radiation. Conventionally, photons are the selected particles for the treatment, but in recent years interest in the use of charged particles has been steadily growing in medical physics. The energy deposition of particle therapy allows to spare the organ at risk around the target volume. At the same time, the energy deposition of particle therapy is greatly influenced by range uncertainties provoked by motion or variations in tissue density along the path, requiring precise treatment planning and monitoring techniques to ensure accurate and effective therapy delivery.One way to achieve this is by analyzing the secondary particles created in the interactions between the charged particles and the tissues. Prompt gammas are candidates for this technique because their creation occurs instantaneously on the interaction site and can escape the patient. For these reasons, prompt gammas are not affected by washout effects and can be easily...
Read More
G.Tec BR41N Hackathon 2023

G.Tec BR41N Hackathon 2023

This weekend 5 of our members (Carola Caivano, Matteo Allione, Matteo Gallo, Marco Casari, Michele Romani) challenged themselves by participating in the hackathon https://www.br41n.io/ organised by g.tec medical engineering GmbH.It was a formative and motivating experience that allowed us to explore the application of CNNs in the field of Brain computer interface, in particular in P300 speller challenge.The results we gained were very accurate and extremely significant and encourage us to carry out a line of research to exploit Transformer models for Causal Language generation, in order to improve the efficiency of spellers for patients who are unable to communicate autonomously. Special thanks for the support to: Pompei Student LabNPO Torino s.r.l.HPC4AI #braincomputerinterfaces#deeplearning ...
Read More
New Machine Learning courses from the MLJC in A.Y. 2022-23

New Machine Learning courses from the MLJC in A.Y. 2022-23

An outstanding opportunity to learn Machine Learning with an "hands-on" approach and get in touch with the Machine Learning Journal Club association. Introduction to Machine Learning Language: Italian Corso articolato in 10 lezioni di circa 2h tenuto presso il Dipartimento di Fisica in orario 16-18.Dopo una prima introduzione al Machine Learning, il corso si focalizzerà sulla comprensione e l'implementazione di algoritmi in linguaggio Python, dai più semplici quali Linear e Logistic Regression fino alle Reti Neurali. Nell'ultima parte del corso verrà introdotto il tema del Continual Learning.Rendiamo disponibili le registrazioni delle lezioni sul nostro canale YouTube.Per iscriversi inviare una email a info (at) mljc (dot) it , le date verranno comunicate man mano agli iscritti. Introduction to ML for Brain Computer Interfaces Language: Italian Corso articolato in 5 lezioni di circa 2h tenuto presso il Dipartimento di Fisica in orario 17-19.Si parte dalle basi dell'analisi di segnali EEG, illustrando varie tecniche e aspetti del preprocessing per arrivare a comprendere e implementare algoritmi di feature extraction e ...
Read More
MLJC mentioned in TG Regione Piemonte

MLJC mentioned in TG Regione Piemonte

On 7th January 2023, RAI TG Regione Piemonte mentioned the Machine Learning Journal Club, its activities and the future Centre for Artificial Intelligence that is expected to be hosted in Turin. https://www.rainews.it/tgr/piemonte/video/2023/01/fermo-al-palo-il-centro-per-lintelligenza-artificiale-f0c4f3cf-320c-462e-a5d5-2d0663710e49.html https://www.rainews.it/tgr/piemonte/notiziari/video/2023/01/TGR-Piemonte-del-07012023-ore-1930-165576ba-cdc6-4766-b5c0-27a1d9267e11.html ...
Read More
Report from NeurIPS 2022, New Orleans, LA (USA)

Report from NeurIPS 2022, New Orleans, LA (USA)

Presentation at NeurIPS 2022, "Symmetries and Geometry in Neural Representations" Workshop (NeurReps) Workshop website | NeurIPS website Last week, some of our members went to New Orleans for the thirty-sixth Conference on Neural Information Processing Systems, one of the biggest gatherings of machine learning researchers from all over the world. Having the chance to bring part of our association to NeurIPS 2022 was a great honour for us. We took part in multiple events. Simone Azeglio and Arianna Di Bernardo co-organised with Sophia Sanborn, Nina Miolane and Christian Shewmake the wonderful Symmetry and Geometry in Neural Representations (NeurReps) workshop gathering together researchers at the intersection of geometric deep learning, applied geometry, and neuroscience (Taco Cohen, Irina Higgins and many more) to study the geometric principles underlying neural representations. Secondly, our members Luca Bottero, Francesco Calisto and Valerio Pagliarino presented their work on learning geometrical features from images through group action enforcement in the poster session (https://openreview.net/forum?id=sEn61s0M1hy).They proposed an autoencoder architecture capable of...
Read More
MLJC project in Geometric DL accepted at NeurIPS 2022, New Orleans (USA)

MLJC project in Geometric DL accepted at NeurIPS 2022, New Orleans (USA)

We are delighted to announce that a group from MLJC including Luca Bottero, Francesco Calisto and Valerio Pagliarino will present their work at the world-leading conference NeurIPS 2022 (Neural Information Processing Systems) in the Neur-Reps (https://www.neurreps.org/) (Symmetries and Geometry in Neural Representations) workshop. The conference will take place in New Orleans, Louisiana (USA) from November, 27th to December 3rd. The project is inserted in the landscape of Geometric Deep Learning (GDL), a relatively recent branch of Machine Learning aiming at studying how to enforce geometrical structures and priors into ML models, whit the final goal of increasing performances, generalization capabilities and explainability. This specific work, titled Unsupervised Learning of Geometrical Features from Images by explicit Group Action Enforcement, has the ability of automatically disentangling the geometric rototranslational and scaling features from the intrinsic ones, when creating a latent representation of a dataset of input images. This line of research may lead in future to the development of more powerful and efficient architectures...
Read More
Presentation of MLJC Projects for A.Y.  2022/23  10th November, Physics Dept. of the University of Turin

Presentation of MLJC Projects for A.Y. 2022/23 10th November, Physics Dept. of the University of Turin

The Machine Learning Journal Club is happy to invite you to the event "MLJC - Presentation of projects and activities for the Academic Year 2022-23". MLJC members will present the key projects and activites currently under development and they will interact with students, researchers and AI enthusiasts for extending our network and opportunities to everyone that is interested in collaborating with us. A large number of highly interactive short talks will allow people to get insight on the technical activities of MLJC in the fields of theoretical and applied Machine Learning, Deep Learning and Artificial Intelligence. Talks will range from Brain Computer Interfaces to ML applications in Physics to Natural Language Processing. Venue: Aula Magna, Physics Department, University of Turin, (Via Giuria 1, Torino)November 10th, from 4 p.m to 7 p.m. Save the dateOpen PDF Recordings Introduction and MedicAI https://www.youtube.com/watch?v=67B-AO4yuss Symmetries https://www.youtube.com/watch?v=Ixv3yzXZkj0 Scientific ML https://www.youtube.com/watch?v=De2eoU3wYYU ML in HEP and more https://www.youtube.com/watch?v=EdSde8p-YVM ...
Read More