User Review( votes)
Artificial Intelligence is not limited to just cancer research, 3D presentation of an object, take facial recognition programs, and observe flora and fauna within the Serengeti. Although, the AI can create and give funny memes also. The two researchers from Stanford University, Abel L. Peirson and E. Meltem Tolunay students from the Physics and Electrical Engineering sections have produced a novel meme-generating method.
They have written a white paper titled ‘Dank Learning: Generating Memes Using Deep Learning’, which is going to produce new and funny memes. Both the scientist have explained a neural network that ingests got the knowledge and dribbles out internet in-jokes. AI consists of the Convolutional Neural Network (CNN) which captures the pictures as input and alter it into mathematical representations called vector embeddings (an encoder), and a Long Short-Term Memory (LSTM) Recurrent Neural Network (RNN) which generate as captions (a decoder).
Both the researchers serve the AI system with 400,000 and more pictures and nearly 2,600 unique picture label-pairs from memegenerator.com. Gadgetsay reporters have found that they train and command AI to especially use “advice animal” style memes which will have a comic caption with a unique character, like a cat wearing a bathrobe. After that, they instruct to examine human problems and had an idea on every picture in a fun way. Now figure out if it is presented by a person or the neural network.
The scientist has stated that it provides for the relatively modest collection of datasets. Here they are clearly assigning to meme generation as work of producing a comic caption which will be related to the real picture. Now it was high time to test the memes both the researchers have to check whether the memes was funny or not. You will be shocked to know that the memes generated from AI got 70% accuracy. The producers had written that the memes produced by human and AI scored close to the hilarity rating. Memes produced from both are tricky to make contrast from a real meme.