Dynamic Compression Flows for Neuroscience Data
Wei, G.*, De Albuquerque, D.*, Martinez, M., Pan, S. & Pearson, J.(2026). "Dynamic Compression Flows for Neuroscience Data." In Proceedings of the 43rd International Conference on Machine Learning.
CV
Education
M.D-Ph.D, Duke University, Aug. 2017-May 2026
PhD in Electrical and Computer Engineering (ECE)
Thesis: Latent Representation Inference for the Biomedical Sciences Through the Lens of Deep Generative ModelingM.S. in Electrical and Computer Engineering (ECE), Duke University, Aug. 2020-Dec. 2022
B.S. in Neuroscience (with Distinction) and Minor in Chemistry, Duke University, Aug. 2012-May 2015
University of Brasilia’s School of Medicine, Brasilia (Brazil), Jan. 2011-July 2012
Completed first three semesters of doctor of medicine (MD) program and transferred to Duke University as an undergraduate on Fall 2012.
Residency & Fellowship Training
Mountain Area Health Education Center (MAHEC), Asheville, NC
Transitional Year Residency Program, July 2026-June 2027University of Washington, Seattle, WA
Diagnostic Radiology Residency Program (Advanced), July 2027-June 2031 (expected)
Honors & Awards
- Selected to serve as Chief Resident, MAHEC Transitional Year Residency Program, 2026-2027
- Invited to serve as a reviewer, Conference on Neural Information Processing Systems (NeurIPS), 2025-2026
- Ruth L. Kirschstein Pre-Doctoral NRSA, NIMH, 2023-2026
- Invited to attend in-person, Innaugural Raynor Cerebellum Project Post-Doc Summit, Scottsdale, AZ, Nov. 2024
- Outstanding Graduate Student Teaching Assistant, Duke University, Spring 2023 & 2024
- Selected to attend in-person, Oxford’s Machine Learning in Medicine Summer School, Oxford, UK, Summer 2023
- Runner-up for Gertrude B. Elion Research Award, Triangle Community Foundation, Spring 2023
- One out of three Duke finalists, Apple’s AI/ML PhD Fellowship selection, Fall 2021
- Medical Scientist Training Program (MSTP) trainee award, Duke University, 2017-2026
- Post-Baccalaureate Intramural Research Fellowsip Award, NIMH, Bethesda, MD, 2015-2017
- Elected Member of Phi Beta Kappa, Duke University, Summer 2015
- Graduated with Distinction in Neuroscience, Duke University, Summer 2015
- Graduated Magna Cum Laude, Duke University, Summer 2015
- Undergraduate Neuroscience Program of Research Fellowship Award, Duke University, Summer 2014
- Outstanding student (top 10% by GPA), University of Brasilia’s School of Medicine (Brazil), June 2012
- Full, merit-based scholarship during all 3 years of high-school, Leonardo da Vinci High School (Brazil), 2007-2010
Research Experience
- Brains & Computation Lab, Duke University, Fall 2019 - May 2026
- PI: John Pearson
- Summary: Research focus was on developing new deep generative modeling approaches better suited to analyze different modalities of neuroscience data (e.g., brain fMRI, calcium imaging data, behavioral data, EEG, ECoG). Projects leveraged general principles of existing deep generative modeling frameworks (e.g., Variational Auto-Encoders, Diffusion-Based Models, Flow-Matching) to create new models capable of inferring latent representations with desirable properties, such as dimensionality reduction, uncertainty quantification, temporal dynamics modeling, and interpretability. Results of this work were published on the Journal of Machine Learning Research (JMLR) and in the proceedings of the 38th Conference on Neural Information Processing Systems (NeurIPS, Vancouver, 2024), and of the 43rd International Conference on Machine Learning (ICML, Seoul, 2026).
- Clinical and Translational Neuroscience Branch (CTNB), National Institutes of Mental Health (NIMH), Fall 2015 - Summer 2017
- PIs: Karen Berman & Joseph Callicott
- Summary: Main project involved combining genome-wide association analysis and neuroimaging methodologies in an agnostic, high-throughput voxel-wise and genome-wide approach to investigate genetic determinants of working memory function. Results of this work were presented at the Society of Biological Psychiatry Annual Meetings of 2016 (Atlanta, poster) and 2017 (San Diego, oral).
- Woldorff Laboratory, Duke University, Spring 2013 - Spring 2015
- PI: Marty Woldorff
- Summary: Main project involved leveraging the high temporal resolution of electroencephalography (EEG) and a previously validated translational sustained attention paradigm to investigate the effects of a visual, global distractor on the temporal cascade of electrical brain processes associated with ongoing sustained attention efforts. Results of this project were first presented in my undergraduate senior thesis dissertation. They also gave origin to an abstract for the Cognitive Neuroscience Society Annual Meeting of 2015 (San Francisco, poster) and are further described in a manuscript published on Neuropsychologia.
Skills
- Programming Languages
- Pytorch, Python, C, C++, SQL, R, Julia, Latex, Markdown
- Spoken Languages
- English (fluent), Portuguese (native), Spanish (advanced), German (beginner)
- Biomedical Data Analysis
- Electroencephalography (EEG) data analysis (ERP and time-frequency methods)
- Neuroimaging (fMRI, DTI, MR spectrostcopy) data analysis using multiple software packages (SPM, FSL, Afni, NiBabel, Pydicom, Nilearn)
- Genomics Data Analysis
Publications
Latent Representation Inference for the Biomedical Sciences Through the Lens of Deep Generative Modeling.
De Albuquerque, D.(2025). "Latent Representation Inference for the Biomedical Sciences Through the Lens of Deep Generative Modeling." Available from ProQuest Dissertations & Theses Global. (3341637296).
Inflationary Flows: Calibrated Bayesian Inference with Diffusion-Based Models
De Albuquerque, D., & Pearson, J.(2024). "Inflationary Flows: Calibrated Bayesian Inference with Diffusion-Based Models." In Proceedings of the 38th Conference on Neural Information Processing Systems.
Deep Generative Analysis for Task-Based Functional MRI Experiments
De Albuquerque, D., Goffinet, J., Wright, R., & Pearson, J. (2021). "Deep Generative Analysis for Task-Based Functional MRI Experiments." Journal of Machine Learning Research. 149(1-29).
The effects of distraction on brain processes underlying signal detection
Demeter, E., DeAlbuquerque, D., & Worldorff, M. (2016). "The Effects of ongoing distraction on the neural processes underlying signal detection." Neuropsychologia. 89(335-343).
Talks & Presentations
Dynamic Compression Flows for Neuroscience Data.
Poster at 43rd International Conference on Machine Learning, Seoul, South Korea
Inflationary Flows: Calibrated Bayesian Inference with Diffusion-Based Models.
Poster at 38th Conference on Neural Information Processing Systems, Vancouver, British Columbia, Canada
Deep Generative Analyses for Task-Based fMRI Experiments.
Poster at Computational and Systems Neuroscience (CoSyNe) 2021 Annual Meeting, Virtual
Genetic Determinants of Working Memory: a High-Throughput Voxel-Wise GWAS Approach.
Talk at Society of Biological Psychiatry (SoBP) 2017 Annnual Meeting, San Diego, California
Assessing Genetic Determinants of Normal Brain Function Through Genome Wide Association Studies (GWASs): a Quantitative Intermediate Phenotype Approach.
Poster at Society of Biological Psychiatry (SoBP) 2016 Annnual Meeting, Atlanta, Georgia
The Effects of Distraction on Brain Processes Underlying Signal Detection
Poster at Cognitive Neuroscience Society 2015 Annual Meeting, San Francisco, California
Teaching
Service & Community Outreach
- Creative Computing Educational Program
- Role: Co-Founder, Instructor
- Active Time: Fall 2022 - Fall 2024
- Description: This program is aimed at teaching 2nd and 3rd graders about basic principles of computer programming logic using Scratch, a free, high-level programming language developed by the MIT Media Lab for educational purposes.
- Boost Beyond Program, Duke University School of Medicine
- Role: Mentor
- Active Time: Fall 2021 - Fall 2023
- Description: This program is aimed at mentoring and supporting high-school students from minority or under-privileged backgrounds interested in pursuing careers in STEM throughout their college application process.
- Hand2Hold Student Group, Duke University School of Medicine
- Role: Executive Board Member; Educator
- Active Time: Fall 2017 - Fall 2018
- Description: This is a student group at the Duke SOM with two main goals: 1) prototyping and constructing customized 3D-printed “wearables” for low-income patients with small hand and/or finger amputations; 2) providing educational outreach to minority students interested in STEM (through Duke’s Boost Program) by introducing these students to basics of hand anatomy, 3D printing, and engineering design.