Neural networks + Kittens = !!!

It was inevitable that the neural network, having named all kinds of internet datasets, should turn its talents to naming cats.

And what an occasion! The AFK Cat Rescue of Huntsville, Alabama contacted me because they had an exceptionally adorable bunch of kittens (plus one magnificent Persian) who need names and homes. June is kitten season here in the USA, so shelters are inundated with new kittens right now, and AFK takes the very high-risk cases, kittens who are too small to survive in regular shelters, or who are sick or injured, or have neurological disorders.

To be sure, the AFK Cat Rescue’s kitten-naming game is strong. They’ve got a black and orange kitten pair named Shere Khan and Bagheera, and they have another pair of cats named Grindylow and Pooka. With their full list of the several hundred cat names they’ve used over the years, plus the list of several thousand cats registered in Toronto (thanks to Tumblr user @maverick-ornithography), I had comfortably enough to train a neural network with.

I fired up my trusty karpathy-char-rnn neural network framework, and sat down to watch it learn.

Colzyy
Mumhan
Tygrar
Juolb Ggonooo Byn
Malgacor
Gatbewl
Mrror
Rglslwelph
Aarla
Teaw Mos Tilypsronvynkor

“Very good,” I said. “Coming right along. Nice work on Aarla, maybe not try so hard on Teaw Mos Tilypsronvynkor.” I said all this to myself, of course, because the neural network operates entirely without my input once training starts. I can shout “No, you fool!” at it all day if I like and it ignores me perfectly.

Soon, however, I began to notice that quite a few of these cats had last names, and sounded actually rather grand.

Jarlag
Argon
Mankith
Cuttim Aeggerooy
Jozga
Andend of Karlans
Irtenda of Tiyra Sittrobt
Torg?
Arten Sword
Lord Magian
Welu-the Manwys
Parihen the Thawk
Haldir of the Saleword Barga
Mr. Yetheract
Belfine Bracken
Belis Goodbrook
Bentone Ballywood
Grim Wyyne
Gorihand
Molgo

I had, as it turned out, accidentally trained the neural network on another dataset, a list of character names from Tolkien, George R. R. Martin, C. S. Lewis, Robert E Howard, and Terry Pratchett (sent in by reader Thomas Pugh).

AFK Cat Rescue, however, decided to roll with it. First neural network kitten: Parihen the Thawk!

Parihen the Thawk: This guy was surrendered to an animal shelter after he hitched a ride in the engine of a car. He had a dislocated leg but it’s healing well with rest. He’s shy, tiny and misses his bff who is in the hospital right now for her much worse injuries. He’s got a lot of energy and loves to show off his belly.

I finally got the neural network training on the proper dataset, but I was worried when, by the time I went to bed it was producing literally the following names over and over:

Hurter
Hurler
Hunty
Hurty
Hunter
Hurker

Some who are not so fond of cats may argue that these are in fact the best cat names. Fortunately for the AFK Cat Rescue, the names did eventually become more suitable. I present to you:

Jexley Pickle - This little girl is a hoot and a half. She’s full of energy, bounce and comedy. She loves to nurse on ear lobes and finger tips. She’s about 6 weeks old and was found after being chased up a tree by a dog.

Mag Jeggles - When you touch him he rattles from head to tail with purrs. He’s so sweet. He was simply too young to be made available for adoption, and was rescued from a shelter that could not care for him.

Snox Boops - He was in a room with a lot of kittens that were too small to place up for adoption and even much smaller than the others in the room. When you pick him up he capsizes in your hands and starts purring. He’s got a huge voice and a huge demand for love and attention.

Mumcake - She was brought into a shelter as a stray, but she was too young for adoption and if she wasn’t pulled by rescue she would have been subject for euthanasia. She’s adorable, loving, outgoing and shoots sunshine right out of her butt. (Sunshine ejection not shown; you have AFK’s word for it.)

Tilly Mapper - This little girl also took a ride in a car engine and her rear leg was nearly completely severed as a result. She’s been stitched up but she’s had an infection set in. She is only around 8 weeks old and she’s in very poor body condition from having tried to make it on the streets. We hospitalized her yesterday and she’s doing much better. We hope to have her back to foster home by Friday.

Big Wiggy Bool - He is a doll-faced persian that was surrendered to a kill shelter when his family could not take him on their cross country move. He’s five years old and has obviously been doted on. He’s super affectionate, very easy going and a talker! He’s been recently groomed and trimmed up by Robyn Warner with Goin to the Dogs and Cats Mobile Grooming Service. And he loves belly rubs. Rubbing that fluffy, soft belly is an experience so wonderful it can only be discussed in breathless whispers.

Photos of the above cats by volunteer Amy Harrell.

And, since it IS kitten season, the neural network is happy to provide a list of cat names (some more usable than others), for use in naming cats, computer servers, firstborn, etc.

Jeckle
Elbent
Jenderina
Roober
Snorp
Snox Boops
Cylon
Sookabear
Frere
Sonney Mrow
Jexley Pickle
Marper
Foppin
Toby Booch
Snowpie
Big Wiggy Bool
Macha Boo
Mr Whinkles
Timble
Macfallon
Machaka
Licky Cat
Mr Bincheh
Macnaw
Maxy Fay
Tim Hike
Mr Gruffles
Grips
Liony Oli
Lingo
Lingley
Conkie
Lasley Goo
Mr Took
Linky
Marvish
Mag Jeggles
Corko
Maggin
Mcguntton
Mara Tatters
Mr Tiggie
Mr. Skuffles
Mr. Hinkles
Mush Jam
Tilly-Mapper
Mr. Jubble
Mumcake
Muppin
Mr O

Perhaps backup choices:

Cutzerinda
Galorub
Pans
Sofa
Shotkie
Ouiho
Pope
Kogon
Ro Larky
Rorka Bot
leperaONtiea
Malool
Scagkaleoru
Clagmlh
Mice
iiia
Ha LuoleryPlogalasnfalon
Hubla Ssrerosti
Negflun Mery
Booii
Balllucidoux

Definitely backup choices:

Trickles
Poot
Moosh Papper
Clotter
Moan
Toot
Cloobie
Slarkbir
Jenky
Pissy
Schitty
Retchion
Pappy
Dopia
Pilly
Scabbys
Pish
Mesladewench
Souffungy
Mr Tinkles