@voronidols are on Twitter, and also on Mastodon @voronidols@botsin.space and the bots page.

Source code

I've updated the algorithm behind neuralgae. It's still the same basic principle: render a deepdraw image based on some random categories, feed the results back through the classifier, and then repeat with the new categories. But the draw part of the cycle now uses the weights from the classification stage …

I've made my first Twitter bot. It posts a new image and accompanying text from my Neuralgae NaNoGenMo project, six times a day, modified slightly so that it fits within the character limit. One of my goals with Neuralgae was to generate something which was in some sense readable, and …

For NaNoGenMo (national novel generation month) 2015, I wrote some Python which generates a series of loosely-coupled deepdream images, and adds eight-line surrealist 'poetry' to them based on the image categories used to draw them.

The full text is here: Neuralgae.

The source code and a fairly detailed description of …

Here are visualisations of the 205 scene categories in the MIT Places Database using a variant of the same technique used to generate the thousand faces of Caffenet.

These are rendered on random grayscale noise rather than a photograph, so they're purely a manifestation of what the neural net has …

Here are the thousand image categories in the ImageNet Challenge data visualisation context, rendered onto a webcam photo of me using the deepdraw technique.

This is the set of image classifications which the standard Deepdreams imagery is based on: as more than one hundred of the categories are dog breeds …

The Deepdreams algorithm takes a deep neural net trained for image-recognition and uses a forced-feedback loop to induce a kind of machine-pareidolia or hallucination, with beautiful and often nightmarish results:

The original algorithm picks details from all of the neural net's training data, based on resemblances to the original image …