Machine learning is the subject which really combines all my interests : on the one hand it is really theoretical and gives me a lot to think about (it is basically modern statistics). On the other hand it is really capable of doing incredible things and can be applied basically everywhere : the only limit is our imagination. In particular, it can be applied in biology and is giving the scientists new and amazing techniques to fight diseases like cancer. Here are some of my projects:
- Here you can find my project report about "Segmentation of neuron bundles from Diffusion MRI", for the course of "Statistical Learning Theory" at ETH Zurich
- Here is my project (and slides) in Image segmentation, for the course "Theory, Programming and
Simulation of Neural Networks" at ETH Zurich. Here I developed a simple histogram clustering algorithm as a block of 2 k-Means clustering blocks. It worked pretty good! Also here you can find another algorithm for supervised learning in Image segmentation using Histograms. I called it "Supervised Histogram Flooding".
- Here is my project about Feature Matching for the Computer Vision course at ETH.
- Here is my project about Camera Calibration for the Computer Vision course at ETH.
Quantum Control (BACHELOR'S THESIS)
I did my bachelor’s degree in the University of Padua (Information Technology Tngineering). My thesis was about the foundations of Quantum Physics, seen as a generalized theory of probability for applications in Quantum Information, Computation and Control. Quantum mechanics is really an interesting topic, however most of the times one has to be a physicist to understand it at a technical level. My introduction is perfect for an engineer!
to download it (thesis is in english, but Slides
are in italian).
For more things I did in Control Theory, check later on this page.
Programming is a lot of fun, here you can find some of my projects.
Test my matrix calculator here!
Here's the code
, if you want to check out how it works!
I am planning to expland this solver implementing eigenvalues calculation using the QR method.
The # of rows and columns must be lower than 9
You can avoid to write zeros
System Identification & Control
Control Thoery is my favorite area of Electrical Engineering, in particular I like the subareas of estimation and identification.
Recently, Control Theory and Machine Learning comminities have been sharing a lot of ideas : I am sure this is going to
matter a lot for future applications in robotics, now that computation power of robots is increasing.
This semester I had the opportunity to follow a beautiful course in System Identification, here are a couple of MATLAB files:
- Here is an exercise on closed loop identification in frequency domain. This file basically contains a summary of all basic closed loop identification methods.
- Here is an exercise on time domain parameter identification in an ARX model.
The thing I love to do the most? spiral around definitions. Definitions are the most important thing for me (yes I love Formal Systems) : when it comes to engineering, sometimes definition are a little bit unclear! Want an example? Check out this
However, some concepts in maths can be pretty breathtaking! Here are some pdfs I wrote about my favorite topics in mathematics.