[Pycon] [new paper] "Cheuk Ho" - AI and algorithmic art

[info a pycon.it]

Title: AI and algorithmic art 
Duration: 45 (includes Q&A)
Q&A Session: 15
Language: en
Type: Talk

Abstract: Abstract

Many people think of neural networks as magic boxes, using them as such is similar to allowing the computer or the machine to have a mind of their own. 
Classification and prediction are the best-known applications of these algorithms. But it has been demonstrated that they can also be used in multiple creative or artistic ways such as generating pictures, musical pieces, humour based dialogues or jokes and plays or literary pieces. Leading to the following question: can these black boxes express true creativity? A trait commonly associated with humans.

Description

Algorithmic Art (or computer generated art) refers to the use of computer algorithms to create artistic pieces. Algorithmic art has been around from the early 1960s, when artists used a plotter controlled by a computer to create some visual artwork pieces. In the 80s when computer graphics became more accessible, digital fractal artworks dominated to become the mainstream of algorithmic art. By the end of the 80s, genetic algorithms had matured enough to have a major influence in the algorithmic composition of music. At the same time, the artificial neural networks were used to explore the creation of musical compositions. 
Most recently, thanks to the blooming of neural network frameworks (e.g. tensorflow), availability of GPUs and development in sophisticated neural network architecture, they play an important part in academic research and data science business applications. Besides that, these enhanced resources and frameworks have enabled the neural networks to make significant contributions in the area of  Algorithmic Arts. Examples of this are the Deep Dream and artistic style transfer and GANs (generative adversarial network) which can generate highly deceptive pictures. A specially trained neural network is also capable of composing music mimicking the style of Beethoven or generating a modern music piece. In the same way, a neural network can be used to generate poems or literary pieces in the true style of Shakespeare or Hamilton.
In this talk, we will go through a gallery of art and music created by algorithms, showcasing what roles computers took in different algorithmic art forms. From the earliest fractal art to the music and pictures generated by the state of the art neural networks and GANs. This talk is suitable for everyone, from the curious general public to experts in the field of neural networks, both will find this talk inspiring and amusing. 

* this talk is a collaboration with Tania Allard

Tags: [u'neural network', u'algorithmicart', u'art', u'pydata', u'Algorithms', u'AI']