Janelle Shane is a user on wandering.shop. You can follow them or interact with them if you have an account anywhere in the fediverse. If you don't, you can sign up here.

Janelle Shane @janellecshane@wandering.shop

The Palmer webcam is looking at its own feet! Looks like fresh footprints leading to the base of the tower. usap.gov/videoclipsandmaps/pal

Just finished Provenance by @ann_leckie and it is SO GOOD! Addendum to the Ancillary Justice series yet pivots to completely different characters, and even completely different *cultures* and it is SO FUN to see what they think of one of the Radchaai when she appears.

realized you could make a sort of markov chain text generator using concatenated word vectors instead of the tokens themselves, which has the benefit of being able to cope pretty well with out-of-vocab strings. anyway, here's word-vector-markov Jane Austen elaborating on what the Internet is

AI-themed burlesque in Seattle Show more

There's going to be a Murderbot NOVEL!! Murderbot is one of my favorite AI characters. If you haven't read the novellas, you are in for a treat. tor.com/2018/07/16/more-murder

Once in a while the Palmer webcam zooms way in on some chunk of the antarctic landscape. This one's particularly nice.

Inept AI attempts to sound sexy Show more

My World article was published today, and I got to include lots of fun anecdotes of gone wrong. physicsworld.com/a/neural-netw

Not sure what this is we found in the woods. Fungus? Dangerous flatsquirrel?

Watching the vigorous bird activity at my parents' feeder, and have decided that nuthatches are my very favorite.
And competition is fierce: cardinals, hummingbirds, red-bellied, downy, AND hairy woodpeckers, tufted titmice, and (close second) carolina wren.
But nuthatches just land on the underside of a tree branch like it's no big deal.

I trained a neural network to invent new names for nail polish. Some of them are very strange indeed.

I trained a neural network to generate new names for fireworks. Here are a few of my favorites.
H/T Bruce Preston for the training dataset of Dutch fireworks
[image: extreme gupperspack & blink, power thunk-7, magical tripe, gold apotionator distroctive XXXXXL 230 stars]

I made a very simple corpus-driven chatbot that replies to you with the line of text from the corpus that comes after/is in response to the line of text most similar to what you just typed. here's a sample interaction where the database is built from the Cornell Movie Dialogs corpus... (my typing in green, bot response in blue; I typed the first turn and it alternates after that)

Hello, everyone! Am writing while sitting near the coffee grinder at this shop. Every time they start it up, the smell is toe-curlingly good.

Via @UKClimbing , the experts weigh in on the future of neural net-named climbing routes.
@McHaffieJames and Visual Chatbot would get along well.

Thanks to UKClimbing.com I had a database of over 427k rock climbing routes to feed to a neural network. It did pretty well. aiweirdness.com/post/175339699

Visual Chatbot doesn't do knock knock jokes.

One fun thing I discovered about Visual Chatbot: no matter what the photo is of, it will always report there are at least a couple giraffes.
It learned from answers that humans gave, and apparently nobody ever asked "how many giraffes are there?" when the answer was zero.

New blog post: When Visual Chatbot is wrong, it just keeps digging itself a hole.
Visual Chatbot is so much fun.