Power-Aware Training for Printed Neuromorphic Circuits

This github repository is for the paper at ICCAD'23 - Power-Aware Training for Printed Neuromorphic Circuits

cite as

Power-Aware Training for Energy-Efficient Printed Neuromorphic Circuits
Zhao, H.; Pal, P.; Hefenbrock, M.; Beigl, M.; Tahoori, M.
2023 International Conference on Computer-Aided Design (ICCAD), October, 2023 IEEE/ACM.

Usage of the code:

1. Training of printed neural networks

$ sh experiment_power.sh

Alternatively, the experiments can be conducted by running command lines in experiment_power.sh separately, e.g.,

$ python3 experiment.py --DATASET 0  --powerestimator power  --powerbalance 0.0   --projectname Power-Aware-Training
$ python3 experiment.py --DATASET 0  --powerestimator power  --powerbalance 0.02  --projectname Power-Aware-Training
...

2. After training printed neural networks, the trained networks are in ./Power-Aware-Training/model/, the log files for training can be found in ./Power-Aware-Training/log/. If there is still files in ./Power-Aware-Training/temp/, you should run the corresponding command line to train the networks further. Note that, each training is limited to 48 hours, you can change this time limitation in configuration.py

3. Evaluation can be done by running the evaluation_power.sh in ./Power-Aware-Training/ folder with

$ sh evaluation_ICCAD_2022.sh

Of course, each line in this file can be run separately as in step 1.