This repository contains a collection of algorithms and models for data analysis and their application to simple tasks. The code was developed during the first half of the course Laboratory of Computational Physics, mod. B, year 21-22.
- Gradient descent: vanilla and stochastig GD, momentum, Nesterov Accelerated Gradient (NAG), RMS Prop, Adam, ADA max.
- Deep neural networks: review of DNN, applications with Keras and Tensorflow.
- Convolutional neural networks: review of CNN, applications with Keras and Tensorflow.
- Combining models, focus on XGBoost and tsfresh: bagging, decision trees, AdaBoost, XGBoost.
- Data visualization and clustering: dimensionality reduction, PCA, t-SNE, K-means, Hierarchical clustering, Density-based (DB) clustering, DBSCAN.
- Unsupervised learning and Boltzmann machines: Restricted Boltzmann machines.
As a weekly exercise, in the second week the professor gave us an assignment on DNN. Unlike the other exercises, in this case we also had to produce a short paper, which can be read at this link.
- Pankaj Mehta et al. “A high-bias, low-variance introduction to Machine Learning for physicists”. In: Physics Reports 810 (2019), 1–124. issn: 0370-1573. doi: 10.1016/j.physrep.2019. 03.001. url: http://dx.doi.org/10.1016/j.physrep.2019.03.001.
- Notebooks of the above review: https://physics.bu.edu/~pankajm/MLnotebooks.html