The MAA is a visionary master program with an innovative and open structure, mixing diverse disciplines, shaping professionals capable of producing theoretical & practical solutions towards responsive cities, architecture & technology.
Reinforcement Learning
Group 5 How can apartment typologies be designed to create communities within buildings, addressing social exclusion and isolation caused by modernist architecture? Can reinforcement learning be used to optimize the distribution of public and private spaces, such as apartments, balconies, patios, and courtyards, to create sub-communities within the building.
Proposal for Enhancement of Semantic Segmentation Model for Architectural Purposes
The current semantic segmentation model for ControlNet provides a list of usable materials, but they are not specific enough for architectural purposes. By generating new tokens (colours) that extend the existing model with new material types, we aim to improve the process by which stable diffusion operates and make it more suitable for architecture applications.