the_decider

This Caravaggio lecture is both funny and informative. His paintings are filled with hidden symbols. https://www.youtube.com/watch?v=kdXVWy6VAwA

I mentioned a “deep” multi-layer neural net, which implies more than just 2-3 layers. As far as I’m aware, this is the first eight-layer trained network mentioned in any publication. http://people.idsia.ch/~juergen/deep-learning-conspiracy.htm...

The first trained multilayer deep neural network was presented by the Soviets in 1970 in a paper titled “Heuristic Self-Organization in Problems of Engineering Cybernetics“ http://www.gmdh.net/articles/history/heuristic.pdf

I personally learn better through coherent code examples rather than math notation / jargon. Manning's excellent "Data Science Bookcamp: Five Python Projects" teaches data science basics using code, not greek symbols. Also, the projects themselves are pretty cool and challenging, but doable. https://www.manning.com/books/data-science-bookcamp

Seems like real-life more closely resembles Kepler’s “Marriage Problem” (later reformulated as the “Secratary Problem). https://www.npr.org/sections/krulwich/2014/05/15/312537965/h... Its a solvable problem from optimal stopping theory. How many people should you date before you know enough to stop and settle (for the next person who’s better than than all the people you’ve dated thus far)? The answer’s x/e, where x can either be a maximum partner quantity, or thr number of years you give yourself to be on the open dating market.

I’m working on a formula that factors ML model run-time cost (speed / memory-usage) into its performance evaluation. I generated a very simple scoring function from economic first principles. Now, I’m trying to test performance on cumbersome transformer models published by Google / FB / Microsoft etc in order to prove that many of these models are not “state-of-the-art” if run-time cost is taken into account.

Agree with this post 100%. There’s too much focus on parameter tweaking, and not enough focus on understanding/ inhancing the data. And feature engineering is considered by many DL practicioners to be“cheating” even though it clearly gets quick results.

Data Science Bookcamp: 10 Python Projects

https://www.manning.com/books/data-science-bookcamp

Those engineers that drafted gas-chamber blueprints had kids to feed too. You’re no better than they are.

This blog-post discusses the dangers of blindly applying English-optimized NLP models to other languages: https://primer.ai/blog/Russian-Natural-Language-Processing/

This book (Data Science Bookcamp) uses actual Python code instead of math notation. https://www.manning.com/books/data-science-bookcamp

Topic modeling of text documents. The so-called LSA topic-modeling technique is basically SVD applied to text. And, as we all know, SVD is simply PCA without data-centralization.

The mind-blowing story behind Ithkuil: It was created by a middle-aged self-educated postal worker. Years later, it became exteremely poplular among Russian Neo-nazis who wanted a philosophically “pure” language for their “pure” race. https://www.newyorker.com/magazine/2012/12/24/utopian-for-be...

Unfortunately, Ferlinghetti's San Francisco is long-gone; the artists have moved on. Tech rules the City now.

Pretty cool extension! I think we all could use a while lot more life and a little bit more color among the drab pages of today’s monolithic web.

Everyone here is too shy to state the obvious; VR poronography will soon be a billion dollar industry that will completely rewire how we as a society seek physical and emotional intimicy. http://www.wired.com/story/coming-attractions-the-rise-of-vr...

Bionformatics in the Bay Area. Afterwards, I switched to Data Science and NLP. Many of my skills were extremely transferable.

When I was post-doc at a well-respected university, the Profs treated me with a combination with disrespect and contempt (when they were not outright ignoring me). I switched to industry; my pay went up 3X-fold, and my talented colleagues praised those problem-solving abilities that I developed over the course of my academic career. My advice to you; f*ck academia! Its a petty pointless prison where you’ll never do anything worthwhile. Switch to industry in order to work and projects that actually have some impact on the world. Only then will you be treated with respect.

I have a Phd in computational biology. Myself and many other computational biologists I know are moving away from the field; transitioning into Data Science, AI, and Natural Language Processing (many of the sequence analytics techniques transfer quite well to NLP). The reason for this Biotech companies treat scientists very poorly, offering them low salaries and a lack of respect. This particularly true for companies like Calico. A friend of mine from grad-school was asked to join Calico during its founding stage. He was promised the world. Instead, he was forced to subsist of a post-doc's salary for years and years. So my advice to you is to stay away from Comp Bio. Consider transitioning into NLP instead.

Cool usage of AI to auto-generate missing wikipedia pages for women of science.

The question is not whether we can have a meaningful conversation, but whether we can have a "functional" conversation. Can the machine quantify a basic statement of my needs, conversationally clarify any fuzzy parameters, and properly execute my request? I believe it can, but only across a set of narrow fields, with much training data and effort.

Browser extensions

Semantic ontologies that would totally change the nature of search.

Now is not the time to compromise; to claim that all opinions equally matter. Now is the time to exclaim that some world-views are good, and some are evil. And that Peter Thiel is not on the side of Good.

The Water issues can be resolved using technologies developed / enhanced by Israel. http://www.haaretz.com/israel-news/science/1.698275

California has access to all the West Coast trade-ports between the US and China. With these ports in place, we don't need to rust-belt states to ship-in goods. An our economy is centered information technology and entertainment, so we don't really need to worry about reduction of economically viable exported goods (all farmed and manufactured products will be consumed internally, thus bolstering that aspect of the California economy).

No. Its a legitimate idea currently making the headway in our community. And if you think its ridiculous, let me ask you...Is the notion of a seperatist, intelligently run Californian sovereign state any less ridiculous than the notion of a Trump Presidency two years ago? We've always been a community of disruptors...well, now's our opportunity to really distrupt.https://techcrunch.com/2016/11/10/calexit-is-silicon-valleys...

Perhaps California's exit from the union will hep YC re-establish its moral integrty. Sam Altman, you wanted to radically change the word (in a positive way)? Well now is your chance!

Under the "Audio" section, we've got Quirious, TalklQ, Twilio; melodious names that end with twirling, soft rhythm. Under "Internal Data", we've got Cycorp and Palantir and Primer; hard-edged P-prominent words implying secrecy and stead-fast solidity.

Some interesting ideas sadly blocked behind a pay-wall journal, all for the purpose of boosting a researcher's prestige because they now hold a "Nature" publication. Thankfully, this article is easily accessible via Sci-Hub. http://www.nature.com.sci-hub.cc/nature/journal/vaop/ncurren...