Most deep neural networks today optimize directly for performance on a task using only example input/output pairs. However, this excludes potential sources of knowledge which could improve performance. We propose a generalized frame-work for neural architectures which contain many semantically-meaningful variables. We call it the Multi-Semantic-Stage Neural Network (MSSNN).
Multi -Semantic-Stage Neural Networks
15.07.2024
1227389 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
Semantic and Logic Theories in Neural Networks
British Library Conference Proceedings | 1994
|River Stage Forecasting Using Artificial Neural Networks
British Library Online Contents | 1998
|Early Stage Surge Discovery Based On Neural Networks
British Library Online Contents | 1996
|