Fork me on GitHub
#data-science
<
2022-01-26
>
Daniel Slutsky10:01:49

Following recent conversations, we are organizing a meetup – hopefully a first on a series – where individuals and groups will share their experiences, problems, hopes, and doubts, regarding the use of Clojure for data problems. https://clojureverse.org/t/real-world-data-meetings/ Your response to the survey will help a lot. Also, please share this with relevant friends.

Daniel Slutsky14:01:04

On Nov. 7th, Mey Beisaron gave a wonderful re:Clojure workshop about Tablecloth. The recording is now finally public: https://youtu.be/VD17eB6vVto

sova-soars-the-sora15:01:00

"unsupervised learning learns patterns from untagged data"

sova-soars-the-sora15:01:41

re: wav2vec it looks like untagged data w/ the right algorithm produces better results than tagged data by a lot. that's so insane. but then, think about if the tags humans ascribe to things are actually helpful for a neural net to deduce patterns.

pavlosmelissinos16:01:10

I remember how amazed I was by word2vec back in the day and at the same time the algorithm was quite simple and easy to understand. Generative adversarial networks and autoencoders use a slightly different approach but have similar aspirations, in case you're interested in learning more about unsupervised algorithms that often work better than supervised ones.

sova-soars-the-sora21:01:44

Well I'm not opposed, whatever it takes to accomplish the goal, especially if it works well, I'm interested in. I would like to start an open-source endeavor for the above ^ (some posts back in #data-science) and I am optimistic that once I start it others will help advance it.