How Data Augmentation Shapes Neural Representations
Tianxiao He, A. Williams, S. Harvey
Data augmentation is widely recognized for improving generalization in deep networks, yet its impact on the geometry of learned representations…
arXiv:2605.15306Statistical Analysis of Neural Data
We develop statistical models and open-source computational tools to extract insights from neural data. We are particularly interested in characterizing flexibility and variability in neural circuits—e.g., how do the dynamics of large neural ensembles change over the course of learning a new skill, during periods of high attention or task engagement, or during development and aging.
Summarizing these processes even on a descriptive level is a difficult and unsolved challenge in need of new analysis tools.
Laboratory for Neural Statistics website
Publications
Tracking the Fidelity of Internal Neural Representations with Error-In-Variables Regression
Isabel Garon, Stephen Keeley, A. Williams
Internal neural representations can systematically deviate from externally measured sensory and behavioral variables, yet neuroscientists lack a principled statistical framework…
bioRxiv:2026.04.22.720005Modulating Cross-Modal Convergence with Single-Stimulus, Intra-Modal Dispersion
Eghbal A. Hosseini, Brian Cheung, Evelina Fedorenko, A. Williams
Neural networks exhibit a remarkable degree of representational convergence across diverse architectures, training objectives, and even data modalities. This convergence…
arXiv:2604.21836
