This file contains the topics the attendants of the Reading Group are interested with. Future weeks papers and exercises will be chosen in order to satisfy this list. Write down any proposal you have.
-
Topics
- Bolztmann Machines.
- Graphical probabilistic models.
- resNet
- Random Forest, XGBoost...
-
Classical papers
- Dynamic Routing between capsules. I have read in a few places already that this might be a very promising future direction for feedforward neural networks. The main idea is that different types of information (shapes, orientations...) are stored in different "capsules" in every layer, each independent of one another, the information in different capsules is sent/re-routed into different capsules in the following layer depending on the information itself. A less-technical description may be found here. It might be very interesting as well to dig in the keras implementation.
-
Quantum papers
- An efficient quantum algorithm for generative machine learning. The title is quite self-explanatory. Indeed, the authors describe a quantum algorithm that allows for creating generative models in an efficient way. It might be interesting to discuss now that we know what a generative model is.