Skip to content
Frontiersin
Search

Lucas Ferreira

Postdoctoral Researcher

Institute for Data-Intensive Science, Brazil

Lucas works on uncertainty quantification for deep learning and its application to scientific discovery pipelines.

Machine LearningData Science
ORCID
0000-0000-0000-0004
Publications
3

Publications

Original ResearchComputational Neuroscience Review

Short-term synaptic plasticity stabilises working memory in sparse spiking networks

Daniel Osei, Lucas Ferreira

Persistent activity has long been proposed as the substrate of working memory, but sustaining it in sparse cortical networks is energetically costly. We show that short-term synaptic facilitation allows information to be held in latent synaptic states, reducing the need for continuous firing. Simulations reproduce the activity-silent dynamics observed in primate prefrontal recordings and predict a characteristic reactivation signature after brief non-specific input.

MethodsApplied Artificial Intelligence Letters

Calibrated uncertainty estimates for machine-learning surrogates in scientific simulation

Lucas Ferreira

Machine-learning surrogates are increasingly used to accelerate expensive physical simulations, but their predictive uncertainty is frequently miscalibrated. We introduce a conformal ensemble procedure that provides finite-sample coverage guarantees without retraining the base model. Across five benchmark simulation tasks, the method achieves nominal coverage with intervals up to 35% narrower than existing baselines.

Original ResearchApplied Artificial Intelligence Letters

Energy-efficient event-based vision with neuromorphic recurrent architectures

Daniel Osei, Lucas Ferreira

Event cameras generate sparse, asynchronous data streams well suited to neuromorphic processing. We propose a recurrent spiking architecture that achieves competitive accuracy on gesture and optical-flow benchmarks while consuming an order of magnitude less energy than conventional convolutional baselines on equivalent hardware.