Search Papers | Poster Sessions | All Posters
Poster A140 in Poster Session A - Tuesday, August 6, 2024, 4:15 – 6:15 pm, Johnson Ice Rink
Toward hierarchical compositionality with shallow hierarchical networks
Francisco Martín López1,2 (), Judith Massmann1, Jochen Triesch1; 1Frankfurt Institute for Advanced Studies, Germany, 2Xidian-FIAS International Joint Research Center, Germany
Unlike conventional deep neural networks, the human brain has a myriad of direct connections from subcortical nuclei to all cortical areas. These enable fast information transfer and facilitate hierarchical compositionality but have not yet been explored in artificial systems. In this work, we present the Shallow Hierarchical Artificial Neural Network (SHANN), a novel brain-inspired architecture with shallow connections from the input to all the hierarchical processing layers. We show that SHANNs can outperform shallow and deep networks in reconstruction and classification tasks. Initial explorations reveal that SHANNs use hierarchical compositionality to combine information from different levels of abstraction.
Keywords: Deep neural networks Hierarchical processing Compositionality Shallow hierarchical networks