Skip to content

Latest commit

 

History

History
47 lines (38 loc) · 1.89 KB

README.md

File metadata and controls

47 lines (38 loc) · 1.89 KB
Raw

Concise description of each paper

The pdf of each paper can be found above.

Master Thesis

This paper is my report for my Master Thesis, done at Harvard University and supervised by Mathieu Salzmann from EPFL It studies error correction in axon segmentation.
The python codes can be found on: https://github.com/AlexandreDiPiazza/Master_Thesis

SatelliteDetection

This paper studies the detection of satellite tracks in images of the universe produced by a satellite. We use Deep Leaning to create synthetic data with a GAN. These images are then used to train a UNET for semantic segmentation.
The python codes can be found on: https://github.com/AlexandreDiPiazza/Semester_Project

OptimizationAlgorithmAnalysis

We study the performance of a second-order optimization method for training a Neural Network, in comparison with the Gradient Descent algorithm.

NNfromScratch

We implemented from scratch a deep learning framework without the use of external libraries i.e coding both the forward pass and the backpropagation algorithm only with basic Python functions.
The python codes can be found on: https://github.com/AlexandreDiPiazza/NNfromScratch

GravitationalLensClassification

We built a CNN to detect gravitational lenses. We reached an accuracy of 91.7%. The model was then used by the lab of Astrophysics at EPFL.

GraphEdgePrediction

This paper presents two algorithms to predict the sign of an edge in a triad of a directed graph. In graph theory, a triad is a subgraph of three nodes representing a triangle. It is studied for example to understand social networks.

BinaryClassification

We implemented different NN models (MLP, CNN, siamese Networks) and compared their performance on a binary classification task.
The python codes can be found on: https://github.com/AlexandreDiPiazza/BinaryClassification