Posted By: OptimusPrime
New NN Architechture - Efficiently Updating Neural Networks - 10/24/20 14:52
Hi Team:
Please take a look at this new NN architecture that is making waves in computer chess circles. Perhaps there is opportunity to integrate this in Zorro. It appears to have wide application.
In 2018, Yu Nasu introduced a new evaluation function based on a shallow neural network whose heavily overparametrized input layer outputs are incrementally updated.
https://www.chessprogramming.org/NNUE
https://www.chessprogramming.org/Yu_Nasu
Thank you so much
Optimus
Please take a look at this new NN architecture that is making waves in computer chess circles. Perhaps there is opportunity to integrate this in Zorro. It appears to have wide application.
In 2018, Yu Nasu introduced a new evaluation function based on a shallow neural network whose heavily overparametrized input layer outputs are incrementally updated.
https:/
https:/
Thank you so much
Optimus