AI Researcher
I am solving operand selection problem for Logic Gate Neural Networks for efficient, explainable and modularizable artificial intelligence.
For now I was able to achieve around 30~40% accuracy for MNIST classification problem with around 1k NANDs. I am now using policy gradient for learning. I am now seeking for other learning algorithms like RL, SAM-like methods to improve the model.
I hope my research could contribute for understanding of knowledge, efficient robotic application and building mutiple-AI system for simulations and computer games etc.