Research output
Peer-reviewed work and current manuscripts in human mobility, geospatial machine learning and multi-label classification. Every record below is rendered directly in the page.
A supervised multi-label framework that identifies concurrent urban functions from mobile-network activity and generalises across Milan and Trento.
A transferable machine-learning framework for detecting cross-border commuting from geolocated social-media traces across three European regions.
An evolutionary approach that deliberately modifies training labels to improve multi-label classification across diverse datasets and models.
Supervised land-use classification from call detail records in an alpine setting, testing generalisation beyond dense metropolitan areas.
A supervised alternative to classical clustering for identifying meaningful places from call detail records.
A locally adaptive method that selects a different neighbourhood size for each prototype in multi-label k-nearest-neighbour learning.
A spatially aware multi-label k-nearest-neighbour method that combines feature-space locality with grid adjacency.