Neural Network Quine

📖 Description

Neural Network Quine Oscar Chang1and Hod Lipson1 1Data Science Institute, Columbia University, New York, NY 10027 oscar.chang, [email removed] Abstract Self-replication is a key aspect of biological life that has been largely overlooked in Arti?cial Intelligence systems. Here we describe how to build and train self-replicating neural net- works. The network replicates itself by learning to output its own weights. The network is designed using a loss func- tion that can be optimized with either gradient-based or non- gradient-based methods. We also describe a method we call regeneration to train the network without explicit optimiza- tion, by injecting the network with predictions of its own pa- rameters. The best solution for a self-replicating network was found by alternating between regeneration and optimization steps. Finally, we describe a design for a self-replicating neu- ral network that can solve an auxiliary task such as MNIST image classi?cation. We observe that th...

📊 Game Impacts

💭 Reactions

🔗 Sources

• [1] https://arxiv.org/abs/1803.05859