Main Works
A collection of highly interesting projects I contributed to.
William Cappelletti
About me
I am a PhD Candidate in Computer Science at EPFL, working at Signal Processing Laboratory LTS4 under the supervision of Pascal Frossard. My thesis focuses on dynamic-graph representation models for multivariate time series.
I graduated from EPFL in Mathematical Engineering, with a Master Thesis on “Byzantine-robust decentralized optimization”, in the Machine Learning and Optimization lab.
I am passionate about Algorithms, Statistical learning and Artificial intelligence. In my free time, I practice skiing, trail running, and rock climbing.
Featured works
-
Learning Self-Supervised Dynamic Networks for Seizure Analysis
Paper, SSL, time series, GNNs, seizure analysis
We propose a GNN model for time series forecasting, that extracts task-specific graphs by self-supervised learning.... -
Byzantine-robust decentralized optimization for Machine Learning
MSc thesis, SGD, decentralized optim., ML
Study on the convergence properties of decentralized SGD variants designed to defend against adversarial attacks -
Polyadic Quantum Classifier
Paper, QML, VQA, NISQ, classification
We introduce a supervised quantum machine learning algorithm for classification on Noisy Intermediate-Scale Quantum architectures. -
Polyadic QML package
Python, quantum, QML, VQA, classification
Python package to define, train and deploy Quantum Machine Learning models through an easy interface. -
Convolutional Neural Tangent Kernel
NTK, Neural Nets, CNN, Kernel regression
In this semester project, I study the relation between artificial neural networks and kernel regression. -
On the volume of zonotopes
Convex Geometry, Zonotopes, Projections, Volume maximization
Study properties of a projection of an hypercube into a smaller space and prove necessary conditions...
Thanks for visting
Do not esistate to contact me.
Until next time!