TED-RNN – Machine generated TED-Talks
Hilarious #ComputationalComedy coding project by Samim. He fed a Recurrent Neural Network with all existing TED Talk Transcripts (1904 speeches or 22.4MB of text which amounts to 4038409 words) and now the model outputs machine generated TED Talks. You can watch three speeches – from Jürgen TEDhuber, Ada LoveTED and Isaac TEDimov – above. Read the transcripts here.
Samim had also created Obama-RNN, a bot that generates political speeches based on a corpus of the speeches of US President Barack Obama. Worth a read.
In addition, I recommend to follow Futuristbot and RNNBible on twitter for more RNN generated content. And have a look at LSTM (Long Short Term Memory), a RNN based on all of the books by Ray Kurzweil, coded by Sebastian Schmieg
[read more] [Samim Twitter] [embedded screenshot by Samin]
(Source: https://www.youtube.com/)