My Research

NeuroModelling

NeuroModelling—the project I submitted most recently (January 2022) to the Spanish State Research Agency (AEI) and that got funded as a Ramón y Cajal grant (started January 2023)—consists in studying the application of spiking neural networks, neuromorphic computing, and quantum machine learning to machine learning problems, in particular to the class optimization problems studied by MODE (see below for my MODE activity).

• Spiking neural networks use, instead of point-like perceptrons, spatially extended neurons where the transport of signals is modelled by differential equations: the increased complexity of these neuron models may help in the optimization of high-dimensional phase spaces.

• Quantum machine learning algorithms are based on encoding information in Qbits (infinite, continuous states) instead of classical bits (binary 0-1 states): the intrinsic differentiability of quantum algorithms, coupled with the quantum advantage, may represent a natural platform for the kind of optimizations we perform in MODE.

MODE

In the MODE Collaboration we are studying the machine-learning assisted optimization of experiment design via differentiable programming. We are currently working on applications of this concept to relatively simple experiment designs, but are already facing computational limitations. I am founding member and steering board member of the Collaboration.

• Muon Tomography Pipeline: with my collaborators, we are currently working on a software package for the automatic optimization of a muon tomography apparatus. The project has applications, among others, in border control, art preservation, and metallurgic industry.

• Neuromorphic computing employs hardware chips based on spatially extended neurons. Such chips are characterized by a lower power consumption than corresponding digital chips. This new class of hardware can drive down the computing resources needed to solve complex optimization problems.

CMS

(Description coming at some point)

• Higgs couplings in CMS
• Machine learning applications in the CMS trigger system