Fully connected networks - 15000 synthesis samples
- Description:
- A synthetic fully connected neural networks dataset, generated with hyperparameters randomly sampled from predefined ranges. The dataset of ~15000 samples contains network configuration and FPGA synthesis data to train regression models for resource and latency estimates.
- Ezzaoui Rahali, Hamza, Mehdi Rahimifar, Mohammad, and Corbeil Therrien, Audrey
- Contributor(s):
- Ezzaoui Rahali, Hamza, Corbeil Therrien, Audrey, and Mehdi Rahimifar, Mohammad
- Source Repository:
- Dataverse de l'Université de Sherbrooke
- Series:
- IMPETUS // rule4ml - Resource utilization and Latency Estimation for ML
- Publisher(s):
- Borealis
- Access:
- Public
- URL:
- https://doi.org/10.5683/SP3/GZNDGW
- Publication date:
- 2024-07-01
- Subjects:
- Keywords:
- Identifier:
- https://doi.org/10.5683/SP3/GZNDGW
Geospatial information
There is no geographic information available for this record.
Select geospatial feature(s) to view on map: