Saheed Faremi
PhD researcher, University College Cork · Data scientist
About
I am a PhD researcher at University College Cork, supervised by Luca Longo at the Artificial Intelligence and Cognitive Load Lab . My doctoral work uses deep learning to study EEG signals. Alongside the doctorate, I build data and AI systems for fintech and healthcare.
Over the past five years, I have built payment systems, fraud-detection tools, document-automation pipelines and machine-learning models for healthcare. I hold an MSc in Computer Science from TU Dublin and a BSc in Information Technology from the University of Eswatini.
Research
Research areas
- Brain and behaviour
- EEG microstates and brain-signal analysis; neuroscientific knowledge discovery; neuroimaging; nonlinear time-series analysis
- Explainable and reproducible AI
- Explainable Artificial Intelligence; mechanistic interpretability; reproducible machine learning
- Applications
- Deep learning for health and behaviour; production AI and data systems
Doctoral research focus
My PhD uses deep learning to test what an EEG microstate actually is, rather than assuming that the classical definition is complete.
- Do brain-state sequences have memory beyond the current state?
- Are brain states truly separate, or are they parts of a continuous flow, and what information is lost when we divide that flow into labels?
- Does the time spent in a state shape the next one, and does the brain’s switching rule stay fixed?
- Which patterns are stable enough to identify a person across sessions and predict cognition or clinical outcomes better than standard statistics?
The work asks which patterns are genuine properties of brain activity and which are created by the model or the way it processes the signal.
Publications
Peer-reviewed
- AICS 2026, University College Dublin (accepted; proceedings to appear) · 2026 · Conference paper
Temporal Dynamics and Latent-Space Interpretability of Generative EEG Microstate Segmentation
AICS 2026BibTeX
@inproceedings{faremi2026temporal, author = {Saheed Faremi and Andrea Visentin and Luca Longo}, title = {Temporal Dynamics and Latent-Space Interpretability of Generative EEG Microstate Segmentation}, booktitle = {AICS 2026, University College Dublin (accepted; proceedings to appear)}, year = {2026}, url = {https://aics2026.com/} } - Brain Informatics · 2026 · Journal article
Integrating Convolutional Variational Autoencoders and the Gaussian Mixture Model for efficient manifold learning and clustering of spatially preserved EEG topographic maps
Brain Informatics · 2026BibTeX
@article{faremi2026integrating, author = {Saheed Faremi and Luca Longo}, title = {Integrating Convolutional Variational Autoencoders and the Gaussian Mixture Model for efficient manifold learning and clustering of spatially preserved EEG topographic maps}, journal = {Brain Informatics}, year = {2026}, doi = {10.1186/s40708-026-00327-9}, url = {https://link.springer.com/article/10.1186/s40708-026-00327-9} } - IEEE ICTAS 2024 · 2024 · Conference paper
Machine Learning Models for Identifying Factors Influencing and Predicting Malaria Among Children Under Five Years in Nigeria
IEEE ICTAS 2024BibTeX
@inproceedings{faremi2024machine, author = {Akinpelumi Saheed Faremi and Boluwaji Akinnuwesi and Elliot Mbunge and Petros M. Mashwama and Stephen Fashoto and Polite Zenzo Ncube and John Batani and Shamsudeen Ademola Sanni and Yinusa A. Faremi and Andile Metfula}, title = {Machine Learning Models for Identifying Factors Influencing and Predicting Malaria Among Children Under Five Years in Nigeria}, booktitle = {IEEE ICTAS 2024}, year = {2024}, doi = {10.1109/ICTAS59620.2024.10507142}, url = {https://www.researchgate.net/publication/380116058_Machine_Learning_Models_for_Identifying_Factors_Influencing_and_Predicting_Malaria_Among_Children_Under_Five_Years_in_Nigeria} }
Preprints, registered protocols & doctoral work
- XAI 2026 (Late-breaking work + Doctoral Consortium track), Fortaleza, Brazil (arXiv preprint) · 2026 · Preprint
BibTeX
@inproceedings{faremi2026interpretable, author = {Saheed Faremi and Andrea Visentin and Luca Longo}, title = {Interpretable EEG Microstate Discovery via Variational Deep Embedding: A Systematic Architecture Search with Multi-Quadrant Evaluation}, booktitle = {XAI 2026 (Late-breaking work + Doctoral Consortium track), Fortaleza, Brazil (arXiv preprint)}, year = {2026}, doi = {10.48550/arXiv.2605.10947}, url = {https://arxiv.org/abs/2605.10947} } - OSF Registries (registered systematic-review protocol) · 2026 · Preprint
Autoencoder-Based Models for Scalp EEG: A Systematic Review of Architectures, Applications, Latent Representations, Interpretability, and Validation
OSF Registries · 2026BibTeX
@misc{akbari2026autoencoderbased, author = {Morteza Akbari and Saheed Akinpelumi Faremi and Luca Longo}, title = {Autoencoder-Based Models for Scalp EEG: A Systematic Review of Architectures, Applications, Latent Representations, Interpretability, and Validation}, howpublished = {OSF Registries (registered systematic-review protocol)}, year = {2026}, doi = {10.17605/osf.io/v2w3z}, url = {https://osf.io/v2w3z} } - XAI World Conference 2025, Doctoral Proposals track · 2025 · Workshop
Explainable Disentangled Representation Learning of Recurring Brain Activation Patterns via Variational Autoencoders
XAI World Conference 2025BibTeX
@inproceedings{faremi2025explainable, author = {Saheed Faremi}, title = {Explainable Disentangled Representation Learning of Recurring Brain Activation Patterns via Variational Autoencoders}, booktitle = {XAI World Conference 2025, Doctoral Proposals track}, year = {2025} }
Under review and manuscripts
- Cognitive Computation (manuscript in preparation) · 2026
Convolutional Variational Deep Embedding Produces Canonical EEG Microstate Maps and Richer Transition Dynamics than Modified K-Means on Resting EEG
Cognitive Computation · 2026BibTeX
@misc{faremi2026convolutional, author = {Saheed Faremi and Andrea Visentin and Luca Longo}, title = {Convolutional Variational Deep Embedding Produces Canonical EEG Microstate Maps and Richer Transition Dynamics than Modified K-Means on Resting EEG}, howpublished = {Cognitive Computation (manuscript in preparation)}, year = {2026} } - ICLR 2027 (under review; OpenReview #50003) · 2026
Spatial Encoding Geometry Determines the Maps and Transitions Recovered by Deep EEG Microstate Models
ICLR 2027 · 2026BibTeX
@misc{al2026spatial, author = {Saheed Faremi et al.}, title = {Spatial Encoding Geometry Determines the Maps and Transitions Recovered by Deep EEG Microstate Models}, howpublished = {ICLR 2027 (under review; OpenReview #50003)}, year = {2026} } - Under review · 2026
Prediction of Early Detection of Autism Spectrum Disorder in Toddlers Using Explainable Artificial Intelligence
Under review · 2026BibTeX
@misc{al2026prediction, author = {Stephen Fashoto et al.}, title = {Prediction of Early Detection of Autism Spectrum Disorder in Toddlers Using Explainable Artificial Intelligence}, howpublished = {Under review}, year = {2026} }
Projects
EEG microstate temporal dynamics
PhD researcher · research
Tests whether including the duration of a brain state improves the model’s match to the recording. With nine states, including duration improves the match from 55.7% to 63.7%, compared with using only the current state. A separate comparison finds that its state-switching patterns differ more from chance than those from standard clustering (35.6% versus 23.6%).
Python · PyTorch · MNE · NumPy · scikit-learn
EEG microstate representation learning
PhD researcher · research
Compares image-based scalp maps with maps built from connections between recording electrodes. When both use the same processing step on 203 resting-state recordings, the electrode-based approach accounts for 61.8% of the signal versus 54.2% for the image-based approach. The image-based approach produces tighter groups of activity.
Python · PyTorch · MNE · NumPy · scikit-learn
EEG microstate architecture search
PhD researcher · research
A 4,832-model architecture search on two high-performance computing clusters, scoring 110 metrics per model and testing consistency across participants.
Python · PyTorch · MNE · scikit-learn · SLURM · IBM Power9
Curnance
Founding engineer · fintech
Multi-asset fintech platform for African markets. Seven production services spanning a double-entry ledger, tiered KYC/KYB, and a fraud/AML intelligence layer.
TypeScript · Go · Flutter · MySQL · Azure Functions · Azure DevOps · Bicep
Talks
Temporal Dynamics and Latent-Space Interpretability of Generative EEG Microstate Segmentation
34th Irish Conference on Artificial Intelligence and Cognitive Science (AICS 2026)
2026
Interpretable EEG Microstate Discovery via Variational Deep Embedding
XAI 2026 (Late-breaking work + Doctoral Consortium track)
2026
-
2024
Experience
PhD researcher
Artificial Intelligence and Cognitive Load Research Lab (AICL), University College Cork
Current role
Uses deep learning to identify recurring patterns in brain activity and to test how long they last and how they change over time.
Data scientist → Founding engineer
Etihuku (Curnance)
2021-05 – Present
Builds data and AI systems for Etihuku clients, including the Curnance multi-asset fintech platform.
Education
PhD
University College Cork
Neural engineering and machine learning
In progress
Supervisor: Luca Longo
MSc
Technological University Dublin
Machine learning, deep learning, data mining, visualisation, statistics
Completed
Completed with a 4.0/4.0 GPA (80/100).
BSc
University of Eswatini
Information technology
Completed
Bachelor of Science in Information Technology, covering machine learning, deep learning, data mining, visualisation and statistics, awarded with Second Class Honours, Upper Division (2.1).