|
Briefing
|
- Music-generating AI – Develops AI program that composes music and sounds human as part of its Magenta program for fostering artificial creativity
- Deep-learning approach – Trained large neural network with tens of thousands of songs, teaching it to predict next note in a sequence
- Follows music rules – Adopted reinforcement learning to introduce simple music rules which improved composition skills compared to other automated music generators
- Implications – Include more creative systems to design products and write eloquent text, while technique of applying rules to systems can be useful in other fields such as robotics, recommendation systems, language translation, and medical diagnosis
|
