Bruno A. Email & Phone Number

Paris, Fr

Greater Paris Metropolitan Region

Laboratoire interdisciplinaire des sciences du numérique (LISN), INRIA TAU, CNRS

São Paulo, Brazil

Title: Learning Structure In Electroencephalogram Using Deep Learning. I'm a PhD student working on Electroencephalogram at NeuroComp Lab, the Federal University of ABC, co-supervised at King's College London. My research interests focus on cognition-based biosignal decoding (e. g. Brain-Computer Interfaces, Emotions, Sleep, and Epilepsy). I am currently working on transfer learning methods in EEG decoding, especially machine learning methods in self-supervised and contrastive learning… Show more Title: Learning Structure In Electroencephalogram Using Deep Learning. I'm a PhD student working on Electroencephalogram at NeuroComp Lab, the Federal University of ABC, co-supervised at King's College London. My research interests focus on cognition-based biosignal decoding (e. g. Brain-Computer Interfaces, Emotions, Sleep, and Epilepsy). I am currently working on transfer learning methods in EEG decoding, especially machine learning methods in self-supervised and contrastive learning paradigms.• Developed a machine learning application using Python for neurophysiological data (EEG, ECG) collection, annotation, visualisation, and experimentation.• Created and normalised physiological datasets to BIDs formats with data structuring and pre-processing for machine learning experiments.• Implemented Pytorch and Scikit-Learn compatible algorithms from the literature, including ShallowNet, Deep4Net, and EEGNet. Trained and fine-tuned deep models using Pytorch and Keras.• Managed the projects using Git and open-sourced the code on GitHub with documentation and tutorials generated with Sphinx. Conducted a literature review of brain signals decoding methods for EEG signals.• Coordinated and published scientific articles. Collaborated with international teams outside my research lab.• Supervised two neuroscience and computer science undergraduate students for one year and half. Instruct and guide them in analysing, discussing and abstracting questions about deep learning methods applied to the brain signal. The focus is on his training as a scientist and a specialist in deep learning. Show less

São Paulo, Brazil

Teaching Assistant in Computer Science Course in Fall 2021, * Computers, Ethics, and Society.• Managed and developed learning objectives for over 200 students. Instructed and guided them in analysing, discussing and abstracting questions about ethics, artificial intelligence and society to form original interpretive arguments. Evaluated and provided critical feedback on over 100 assignments.

Greater Paris Metropolitan Region

Laboratoire interdisciplinaire des sciences du numérique (LISN), INRIA TAU, CNRS

Greater London, England, United Kingdom

University of Glasgow and Getúlio Vargas Foundation.I worked on a project to automatically crawl tweets from the largest city in Brazil and associate them with official flood indices. Building and structuring an SQL database with flood related tweets. This work is part of the interdisciplinary project Waterproofing Data, coordinated by Prof Dr Joao Albuquerque.

São Paulo, Brazil

• Developed a crawling application using Python for real-time data collection, annotation, visualization in a massive volume of data using Waze data stream in the largest city in Brazil, São Paulo (using AWS);• Used distributed computation for calculated metrics related to a delimited street of interest of the client, the Traffic Engineering Company, metrics such as the delay on main avenues of the city or accident notified with high confidence in an area.• Integrated the official… Show more • Developed a crawling application using Python for real-time data collection, annotation, visualization in a massive volume of data using Waze data stream in the largest city in Brazil, São Paulo (using AWS);• Used distributed computation for calculated metrics related to a delimited street of interest of the client, the Traffic Engineering Company, metrics such as the delay on main avenues of the city or accident notified with high confidence in an area.• Integrated the official accident database with police and hospital notes on the severity of vehicle accidents for the application.• Designed experimental protocol to aggregate heterogeneous databases, with some variables in one that don’t exist in the other, to associate the Waze Data with the municipal vehicle accident database with terabytes per days.• Extracted technical and business requirements with the client and other actors involved. Participated in several meetings to enable the implementation of the Minimum Viable Product developed.• Implemented and used machine learning algorithms including CatBoost, XGBoost, LightGBM, Random Forest and other decision trees to relate the Waze data with the city’s official car accidents database.• Deployed it within a system used for prediction on large volumes of data, with a re-training schedule to determine accident severity. Used pre-trained AI models from the scikit-learn and CatBoost packages.• Built a dash dashboard to present each information constructed in a responsive behaviour architecture for various web platforms, with an intuitive design and user control of parameters.• Managed the project using Git and open-sourced the code in GitHub with documentation.The goal of this project was to use Waze data to determine if an accident alert was critical or not. The project was funding by the Inter-American Development Bank (IDB). This innovative pilot project focus on massive data governance in four cities and a district in Latin America. Show less

São Paulo, Brazil

• I was an analyst in the risk team of the largest bank in Latin America, market values US$ 39.06 billion in 2022. Responsible for internal evaluation for proxy communications, firewall, and sustained automation requests until verification of compliance with the General Data Protection Law - LGPD.• Responsible for process improvement, internally evaluated requests with natural language processing models. I used Python to pre-classify internal requests and reduce their processing time.•… Show more • I was an analyst in the risk team of the largest bank in Latin America, market values US$ 39.06 billion in 2022. Responsible for internal evaluation for proxy communications, firewall, and sustained automation requests until verification of compliance with the General Data Protection Law - LGPD.• Responsible for process improvement, internally evaluated requests with natural language processing models. I used Python to pre-classify internal requests and reduce their processing time.• I was the second-fastest analyst in demand resolution during this period.I left the position to start the PhD. Show less

São Paulo, Brazil

• Led a pilot data science project and a new product to build stock market volatility indicators based on network topology from media news outlets (Brazilian Stock: IBOV and WDO). • Integrated a machine learning model into a media news crawl built inside the Google Cloud Platform, Big Query from Google News API. • Extracted measures from the complex network and ensemble with a cluster to create a cluster of news.• Developed custom pre-processing and normalising unstructured… Show more • Led a pilot data science project and a new product to build stock market volatility indicators based on network topology from media news outlets (Brazilian Stock: IBOV and WDO). • Integrated a machine learning model into a media news crawl built inside the Google Cloud Platform, Big Query from Google News API. • Extracted measures from the complex network and ensemble with a cluster to create a cluster of news.• Developed custom pre-processing and normalising unstructured information, using natural language processing with Stemming, Lemmatization, and other regularisation methods to normalise different text sources.• Created and drew an architecture based on a complex network, considering the similarity of each pair of news to build edges and vertices with the Latent Dirichlet Allocation model.• Extracted measures from the complex network (PageRank, K-core, small world parameter) and ensemble with a natural language cluster to enable a cluster of news.• Monitored select metrics in the sub-cluster based on a non-parametric statistical test, and I applied tests on the distributions to verify correspondence as a thermometer of political instability.• Interacted and interviewed end-users to identify problems and triggers that could be solved by machine learning and data analysis.• Preliminary results showed this methodology could be used to decrease money loss by 10%. Show less

Santo André, São Paulo, Brasil

I was a three times fellowship student at the Federal University of ABC funded by UFABC and CNPq (Brazilian National Council for Scientific and Technological Development). I worked on three projects:In the third year, I work with Data Analysis to Support Undergraduate Decision Making.

São Bernardo Do Campo, São Paulo, Brasil

• Preprocessed the EEG using EEGLab with ICA and band-decomposition (alpha, beta, gamma, theta) in Matlab. • Constructed the epoch from the continuous biosignal related with a motor imagery experiment with data from three hospitals with chronic pain patients and a control group.• Designed and Implemented a feature learning pipeline to extract measures based on functional and effective connectivity estimation methods for Electroencephalography.• Extracted complex network measures of… Show more • Preprocessed the EEG using EEGLab with ICA and band-decomposition (alpha, beta, gamma, theta) in Matlab. • Constructed the epoch from the continuous biosignal related with a motor imagery experiment with data from three hospitals with chronic pain patients and a control group.• Designed and Implemented a feature learning pipeline to extract measures based on functional and effective connectivity estimation methods for Electroencephalography.• Extracted complex network measures of functional and effective connectivity, such as small worldness, cluster coefficient and average path length across brain network clusters.• Validated the findings of the differences in functional brain connectivity and discovered a modulatory behaviour in specific brain oscillations with a non-parametric statistical test (Kruskal–Wallis with Friedman).• Conducted literature review of the method Cross Convergent Mapping - CCM, an extracted of casualty and a statistical approach based on the theory of dynamical systems and mental pain processing.• Presented at monthly journal club meetings for a research group in Neuroscience. Lead discussions regarding novel trends in the field, article critique and possible directions for improvement. Focused on cognition, pain and statistic.• Achieved statistic significance inter-groups, chronic pain patients and control group, during the imagery motor task related to the source of the chronic pain.Summary, I applied Computational Neuroscience on Brain Functional Connectivity with Causality in Time Series. The project aimed to distinguish if brain connectivity in chronic pain patients was affected. For this, I used a feature learning pipeline to extract the connectivity in Electroencephalography. Show less

São Paulo, Brazil

In the first year, the research project was on One-Classification Thematic (also known as anomaly detection) with traditional machine learning and Complex Networks methods.

Campo Grande

I received an internship fellowship to be a research intern at Embrapa Beef Cattle, under the supervision of Prof. Edson Takashi Matsubara from Artificial Intelligence Laboratory at FACOM/UFMS. During the internship, the goal was to select attributes with a good explanation of the trough behavior of the animals and their thermal comfort.

Campo Grande E Região, Brasil

I was a high school researcher at the computer vision laboratory. I worked for one year and a half in a research group at Inovisão/UCDB. In conjunction with two undergraduate researchers, the project analyzed the behavior of bamboo weevils during feeding acceptance experiments.I developed two papers: Bamboo Weevil Counting Using Threshold Techniques, presented at Workshop of Undergraduate Works - Conference on Graphics, Patterns and Images - Sibgrapi (2015), and "Rastreamento do… Show more I was a high school researcher at the computer vision laboratory. I worked for one year and a half in a research group at Inovisão/UCDB. In conjunction with two undergraduate researchers, the project analyzed the behavior of bamboo weevils during feeding acceptance experiments.I developed two papers: Bamboo Weevil Counting Using Threshold Techniques, presented at Workshop of Undergraduate Works - Conference on Graphics, Patterns and Images - Sibgrapi (2015), and "Rastreamento do caruncho do bambu usando fluxo óptico", presented at Computer on the Beach (2015). Show less

Doctor Of Philosophy - Phd, Computer Science

Activities and Societies: Cotutella with UFABC

Doctor Of Philosophy - Phd, Computer Science

Deep Learning in Electroencephalogram. sup: Raphael Camargo and Walter Pinaya.

Bachelor'S Degree, Neuroscience

Functional Connectivity Study in Electroencephalogram;

Bachelor'S Degree, Computer Science

Activities and Societies: Course Student Representative (2018-2019), Dissertation in Data Science in Academic Learning. Relevant.

Bachelor Of Science - Bs, Science, Technology And Society

Technician Of Computer, Computer Science

Activities and Societies: High School Scientific Initiation fellows (3 years), Co-founder of the student union. Six-time Medalist.