Description
We explore the concept of "space folding" in relation to neural networks. Invited researcher Michal Lewandowski is an expert in this field, and the aim of the research kitchen is to explore how this concept can be beneficial for the research of the participants.
Short on space folding (from an abstract by Lewandowski): Recent findings suggest that consecutive layers of neural networks with the ReLU activation function fold the input space during the learning process. While many works hint at this phenomenon, an approach to quantify the folding was only recently proposed by means of a space folding measure based on the Hamming distance in the discrete activation space.
Invited researcher
- Michal Lewandowski, Researcher and Senior Data Scientist, Software Competence Center Hagenberg GmbH (SCCH)
Programme
Upon request
Practical
Attendance by invitation only.
About research kitchens
A research kitchen is a small, collaborative workshop designed to bring together a group of researchers with a shared interest. The aim is for participants to meet and “cook” interesting “dishes” together. These dishes may take the form of new ideas, collaborations, topics, results, questions, papers, project proposals for funding, and more. All participants are expected to bring their own “ingredients” to the table and to actively engage in the process.
Organisers
- PhD Fellow Johan Mylius Kroken
- Integreat Administrative Leader Maria Dikova, mariyad@math.uio.no