Google Brain has implemented a neural network capable of encryption and decryption.

Oct 31, 201635

AI Translation

This post is translated from Chinese into English through AI.View Original

AI-generated summary

Google Brain has developed a neural network that can encrypt and decrypt information. The network involves three roles - Alice, Bob, and Eve - with each role represented by a neural network. Through a generative adversarial network, Alice and Bob can communicate securely while Eve is unable to access the information. This example demonstrates the importance of mutual learning and growth. The experiment was conducted using TensorFlow on a single GPU.

Google Brain has done something quite crazy again.

Using three roles to create an encryption and decryption machine learning system - Alice, Bob, Eve.

Alice and Bob need to communicate with each other, while Eve tries to steal data from them.

These three roles are neural networks.

Alice: The loss function is such that the more accurate Eve's guesses are compared to random guessing, the greater the loss.

Bob: The loss function increases as the distance of the guess grows.

Eve: The loss function decreases as the guess becomes more accurate.

This forms a generative adversarial network, allowing Alice and Bob to communicate with each other, creating an encryption and decryption network, and preventing Eve from obtaining the correct plaintext information.

This is an important example of mutual learning leading to mutual growth.

The paper in the link mentions experiments done using TensorFlow on a single GPU: https://techcrunch.com/2016/10/28/googles-ai-creates-its-own-inhuman-encryption/