Algorithmic Traffic Routing
Petar discusses the challenges of applying algorithms like Dijkstra's to real-world traffic data, highlighting the need to bridge the gap between raw data and algorithm inputs. He introduces the concept of using a high dimensional graphnet pre-trained to execute Dijkstra's algorithm in a latent space, enabling more efficient mapping of complex, noisy real-world data for traffic routing.In this clip
From this podcast
Machine Learning Street Talk (MLST)
#85 Dr. Petar Veličković (Deepmind) - Categories, Graphs, Reasoning [NEURIPS22 UNPLUGGED]
Related Questions