8 papers
Evaluating Deep Multivariate Imputation Models on Wearable Device Data
Skye Goodman, Roussel Desmond Nzoyem, Leandro Junges +5
Wearable device data enables continuous health monitoring, but suffers from structured missingness: features sharing a physical sensor drop out together. Deep imputation methods su…
Render, Don't Decode: Weight-Space World Models with Latent Structural Disentanglement
Roussel Desmond Nzoyem, Mauro Comi
Training world models on vast quantities of unlabelled videos is a critical step toward fully autonomous intelligence. However, the prevailing paradigm of encoding raw pixels into…
Out-of-Support Generalisation via Weight-Space Sequence Modelling
Roussel Desmond Nzoyem
As breakthroughs in deep learning transform key industries, models are increasingly required to extrapolate on datapoints found outside the range of the training set, a challenge w…
FLEX: Feature Importance from Layered Counterfactual Explanations
Nawid Keshtmand, Roussel Desmond Nzoyem, Jeffrey Nicholas Clark
Machine learning models achieve state-of-the-art performance across domains, yet their lack of interpretability limits safe deployment in high-stakes settings. Counterfactual expla…
Language Models Do Not Embed Numbers Continuously
Alex O. Davies, Roussel Nzoyem, Nirav Ajmeri +1
Recent research has extensively studied how large language models manipulate integers in specific arithmetic tasks, and on a more fundamental level, how they represent numeric valu…
Weight-Space Linear Recurrent Neural Networks
Roussel Desmond Nzoyem, Nawid Keshtmand, Enrique Crespo Fernandez +4
We introduce WARP (Weight-space Adaptive Recurrent Prediction), a simple yet powerful model that unifies weight-space learning with linear recurrence to redefine sequence modeling.…
Towards Foundational Models for Dynamical System Reconstruction: Hierarchical Meta-Learning via Mixture of Experts
Roussel Desmond Nzoyem, Grant Stevens, Amarpal Sahota +2
As foundational models reshape scientific discovery, a bottleneck persists in dynamical system reconstruction (DSR): the ability to learn across system hierarchies. Many meta-learn…
Reevaluating Meta-Learning Optimization Algorithms Through Contextual Self-Modulation
Roussel Desmond Nzoyem, David A. W. Barton, Tom Deakin
Contextual Self-Modulation (CSM) (Nzoyem et al., 2025) is a potent regularization mechanism for Neural Context Flows (NCFs) which demonstrates powerful meta-learning on physical sy…