An efficient technique for optimising tuning curves starting from behaviour by iteratively refitting the tuning curves and redecoding the latent variables.

We use an unsupervised sequence detection algorithm to demonstrate, for the first time, replay of procedural memories in the striatum and that this replay is hippocampus-independent.

A model of the hippocampal formation where theta-band oscillations and local error-correcting Hebbian learning rules train weights it analagous to a wake-speel algorithm. This bioplausible hippocampal model replicates key cognitive functions including path integration.

This paper documents and motivates the RatInABox package, used for modellign locomotion and neural activity.

This short paper demonstrates a symmetry between theta sequences and eligibility traces, potentially explaining how credit assignment is implemented in the brain.

We study how successor representations can be learned in a biologically plausible way by hippocampal neurons using STDP and theta phase procession.

Convolutional neural networks are used to solve the outstanding problem of heat flux prediction in turbulent oceanic flows.

We built a reservoir network architecture, based off Asabuki et al. (2018), which can chunk sequential data into meaningful segments without supervision.

We train neural networks to solve multiple tasks simultaneously and study when and why individual neurons become selective or mixed-selective.

We use variational autoencoders to compressed ovarian cancer transcriptomes and extract genetic markers predictive of chemotherapy resistance.