Niels Verleysen's Projects
A summarization model finetuned for research papers
Following along an older course on RL, this code is based on deprecated libraries.
Sample application for a CI workshop using free tools for open source projects.
Repository following along with my blog series about deep learning.
Following along the course Deployment of Machine Learning Models (https://www.udemy.com/course/deployment-of-machine-learning-models/)
Class that can be used to export the time series from a selection of the sensors from a given satellite.
An example of using huggingface to finetune an image captioning model
Experiment for my thesis. In this repository I will test the predictive quality of an LSTM-network on a periodic function with different sampling rates, types of noise and amount of previous values given as input.
My notebooks for following along the course neural networks from scratch.
Proximal Policy Optimization from scratch with PyTorch for gym environments
Building an object detector (YOLO) while learning PyTorch
Different Q-learning algorithms tested on OpenAI Gym
Following along the course on Deep Reinforcement Learning by Huggingface (hosted here: https://huggingface.co/deep-rl-course/unit0/introduction?fw=pt)
This is the implementation of my research project for the Multi-Agent Systems course at KU Leuven. In this project I reproduced a self assembling robot swarm in the MASON simulation environment. The system gets a pattern as input and the agents move to reconstruct this pattern.
Some small programs showcasing computer vision algorithms
Main experiment of my thesis: Temporal modeling for safer spacecrafts. In this repository I use an LSTM network for forecasting of multiple time series coming from spacecraft sensors. I am using a subset of sensors from the Mars Express orbiter.
Repository to play around with different RNN architectures for my thesis on temporal modeling for safer spacecrafts. I'm using python and tensorflow/keras.
Implementing the transformer architecture from scratch with pytorch
Playground to develop game playing agents for the VerhArcade machine at our Verhaert offices