Nicola Muça Cirone
Postgraduate Researcher
Imperial College London
I am currently pursuing a PhD at the Oxford-Imperial CDT in Mathematics of Random Systems, under the guidance of Dr. Cristopher Salvi at Imperial College London. I completed my undergraduate studies at Università di Pisa in 2022, with my thesis conducted at ETH Zürich under the supervision of Professor Josef Teichmann.
My research interests are centered around time series analysis within the fields of Machine Learning and Mathematical Finance. I approach these areas through Rough Paths Theory, as well as Functional and Stochastic Analysis.
Currently, my work is focused on exploring the scaling laws that govern the behavior of randomly initialized Neural Networks, both pre-training and during training.
I am particularly interested in Neural Controlled Differential Equations and State Space Models.
From February to March 2025, I will be interning at Cartesia.ai as a PhD Researcher.
Reviewed Publications
Muça Cirone, Lemercier, Salvi. “Neural signature kernels as infinite-width-depth-limits of controlled ResNets”, ICML23.
Muça Cirone, Orvieto, Walker, Salvi, Lyons. "Theoretical Foundations of Deep Selective State-Space Models", NeurIPS24.
Preprints
Muça Cirone, Hamdan, Salvi. "Graph Expansions of Deep Neural Networks and their Universal Scaling Limits", August 2024.
Muça Cirone, Salvi. "Rough kernel hedging", January 2025.
Glückstad, Muça Cirone, Teichmann. "Signature Reconstruction from Randomized Signatures", February 2025.
Muça Cirone, Salvi. "ParallelFlow: Parallelizing Linear Transformers via Flow Discretization", February 2025.