Yves Ruffenach

Research profile

Yves Ruffenach develops the EVE variational neuron and studies how local probabilistic computation can make uncertainty measurable inside neural networks and language models. The research program covers the computational primitive, its operating regime, Transformer integration, reliability monitoring and uncertainty-aware control.

ORCID 0009-0009-4737-0555 · arXiv author search · Email

Publications

1. Variational Neurons and Bayesian Neural Networks: Distinct Probabilistic Objects, Inference Loci, and Computational Granularity (August 10, 2026 · Zenodo · Preprint)
2. Distributional Neurons: Making Uncertainty a Unit of Computation (2026 · Research Square · Research Article)
3. The Neuron Is the Distribution (July 14, 2026 · Research Square · Research Article)
4. The Neuron as a Latent State: Classical Variational Readout in Distributional Neural Units (June 30, 2026 · Research Square · Research Article)
5. Measuring and Controlling Internal Activity in Variational Neural Units (May 4, 2026 · Research Square · Method Article)
6. Variational Distributional Neuron (2026-02-20)
7. Exploring the Dimensions of a Variational Neuron (2026-03-14)
8. Variational Neurons in Transformers for Language Modeling (2026-03-30)
9. Agentic Control in Variational Language Models (2026-04-14)
10. Learning Distributions Inside a Language Model: Variational Neurons for Measurable Internal Uncertainty and Reliability Monitoring (2026-07-27)
11. Measuring Internal Probabilistic Activity with Variational Distributional Neurons (2026-07-29)
12. Measuring Local Posterior Activity in Variational Language Model Units (2026-07-28)
13. Variational Distributional Neurons for Measurable Internal Uncertainty in Language Models (2026-07-29)

Research topics

• Variational neuron
• Variational language models
• Variational Transformer
• Internal uncertainty
• Variational closure