Marta Blanco-Pozo

A geometric and dynamical theory of latent computations in biological neural networks

The paper introduces latent processing units (LPUs) to explain how neural networks perform reliable computations despite variability, using six theorems to unify geometric and dynamical perspectives.

Low-dimensional Neural Codes Suppress Neuronal Noise and Extend the Working Memory Duration

The paper shows that low-dimensional neural coding suppresses noise and extends working memory duration in recurrent networks through theoretical analysis and experimental validation in mice.

Building Goal-Directed Cognitive Graphs

The paper introduces a reinforcement-learning framework called Sparse Cognitive Graph (SCG) that models goal-directed behavior through dynamic graph reconfiguration, showing it explains diverse behavioral regimes and predicts population activity geometry.

ActSort: An active-learning accelerated cell sorting algorithm for large-scale calcium imaging datasets

ActSort reduces human labor in cell sorting for calcium imaging by 1-3% and improves accuracy through active learning and expert features.

Dopamine-independent state inference mediates expert reward guided decision making

The paper shows that dopamine-independent state inference and dopamine-mediated reinforcement learning interact to guide reward-based decisions in mice.