What does Veeky Forums think about this paper I published with my undergrad research advisor Chris Snow...

I like you too
I don't have one, I bailed out of academia entirely with my bachelor's degree and went to Wall Street

You could say I met my match in Chris Snow and realized I could never match his higher native intelligence and more years of being enthusiastic about his particular field than I've even been alive

I am thankful to Caltech for mercilessly exposing me to this un-PC reality before I'd made the mistake of thinking about which grad school I should apply to

This trailer kind of reminds me of what these SURFs were like

Stop bumping your own thread you autistic shit stain, why are you flaunting a paper from 9 years ago

I'm just trying to show off my paper for fuck's sake. Do you know there are actual PhDs on this site?

It's just one little undergrad research project paper. I don't see why any PhDs would be afraid of it. None of the PhDs were afraid of it at the time, so what's the problem?

If there's one person you should be afraid of, it's the collective loli known as the Arnold Lab at Caltech

Not to be confused with the Baker Lab, which is the collective loli from whom I stole all my ideas for the benefit of the Arnold Lab

In conclusion, everyone should contribute spare CPU cycles to the Baker Lab's Folding@Home project.

Aren't academic citations 100x more fun when they're handled by software and you don't have to go up on stage looking like an ass saying "no, *I* did that"?

In a legal sense, it's kind of like double jeopardy. Why would anyone have to defend, much less a PhD thesis, some random paper they've already published, all over again from first principles?

You could even imagine a machine learning AI that defends a random paper from its own citation tree from first principles, but that's a PhD thesis all in its own

I know it has to do only its own citation tree because in the general case for any paper, that would be reducible to some kind of superintelligent science AGI.

If it's only the citation tree, you'd need a full learnable model of academic peer review, but that's not even a PhD level problem

However, for the record, it's also not really an undergraduate level problem, which leaves the poor SURFs in an awkward position

Now Yaser Abu Mostafa is surely having second thoughts about his YouTube CS 156 course, but in what possible alternate universe could I have had any coherent complaint about this course at any time, in the past or foreseeable future?

I know I have been making a lot of incoherent complaints about a lot of thibgs ever since I took this course, but never about the course itself

This also explains my time at Virtu. I had a lot of incoherent complaints in my head about most of the things anyone ever said to me, but I was never able to express them coherently

forgot my pic

Yaser Abu Mostafa said to me, not directly but through his TAs "I'm not sure I understand what part it is you don't understand". I replied, "I don't know either, but here's my Netflix challenge thing with SVMs or whatever".

Much like all my other graded CS courses, I got an A in this class even though my actual individual grades on the homework and finals before the curve were an F-

I remember being a TA in CS 1a and giving all my students an average of 0, which must be an F-- so this is probably fair

In English, an F-- would be, not only did d you fail, but all your friends did too, just not quite as badly as you did

In any other school, this would sound like a serious accusation of cheating, but Caltech has a strict honor code that I have mostly never violated and mostly never seen anyone else violating

How have I violated the honor code? This is a true Caltech confession. I have never collaborated with anyone against the honor code, but on multiple, escalating occasions I have taken more time than I was supposed to, and even looked up theorems and proofs in math papers on the internet. Don't try this at home, kids.

Who knows, they might even have to lower my GPA from an A to a D for this, like it was before the curve, and take away my upper class merit scholarship. Thank God I've already graduated!

But what if it's all the way to an F, and not just a D? That's for Caltech administration to somehow decide. This was like 10 years ago after all.

I knew those 80% of Caltech courses where I was failing hilariously but switched to P/F on drop day and ended up passing had to be good for something

You can tell which courses I cheated on the least by how badly I was failing, except for that fucking intro number theory course where I tried to cheat as hard as I could but still almost failed until I figured out how to prove like one of the theorems

So from a formal perspective, this is one class in which I think (IIRC, if I even passed it and didn't drop it on drop day entirely) where I got a P but definitely deserved an F. But then you'd have to reevaluate all my other Ps as well, and those cases are much less clear

There was no penalty for this behavior in any of my classes at the time, but it was certainly known to be, and turned out to be in actual fact, very bad karma

Fortunately for me, ever since then I have been, at worst, involuntarily psychiatrically hospitalized 12 times by my retarded mother Alfia and my retarded ex Caltech gf Lauren, so how much more can I even expect for cheating a few times in college?

Maybe it's been Richard Feynman's ghost that's been haunting me this whole time, bearing an indelible grudge for my getting an unfair undergrad Caltech grade

In all serious though, it would actually be a grave insult to Richard Feynman himself if nobody corrected both my grade and my hospitalizations.

If Richard Feynman himself were to inspect all my Caltech homework, what do I think he would find? A lot of misguided nonsense in bad handwriting, a few good ideas here and there, and a lot of Ps instead of Fs

Better to have Feynman say to you "this entire line of reasoning is complete nonsense, and here's why", rather than "I literally don't even know what you're talking about in this specific sequence of English words in your proof"

This thread has been quarantined to control radioactive levels of autistic same-phag blogging.

I guess the only way for an idea to be so completely wrong that even an undergrad TA can prove it wrong is for the proof itself to be blatantly self-contradictory. Otherwise you might need to call in a PhD

How can an undergrad write an entire pages long proof that is blatantly self contradictory from line 1 and still feel good about it? It's because he or she is not a PhD

The worst case is when it's absolutely right on lines 1-10, wrong on lines 13, 57, and 102, and you have to explain in the margin in red pen which particular bag of unholy CS or ML theorems actually proves this wrong

Thank God in my SHARPEN paper there was only some kind of jerry rigged type theory and distributed computing stuff and no actual machine learning involved. We didn't do any machine learning at Virtu either the whole time I was there

I was writing in my journal recently about a 2D or 3D vector/video/video based symbolic human readable language based on cons cells and machine learning, but I never got anywhere with this on my own

I was thinking about a language that with backreferences could automatically look things up for me and rewrite things in my own words using machine generated cons graphs, but this was impossible for me. I was using the Python library moinmoin and vim

You can even do local, static, or dynamic time varying cons hypergraphs in N to infinite dimensions. There is no limit to what you can do with category theory, which is also something I don't know and have never studied

Or global, if you dare. But global is only relative to your current physical machine and network topology, process ID, and symbol table

This language would ideally be JIT compiled on the fly as you type, and not interpreted

In SHARPEN, this would probably be written as

> import Sharpen.Distributed.MachineLearning.JIT
instead of
> import Sharpen

With a custom Python import module to resolve the hierarchy of the submodules

There is even, if you are prepared to truly lose all your sanity and reason, an exception that the import module throws if it can't find a module, that you can feed directly back into the machine learning distributed JIT compiler

You could even make an entire hierarchy of multiclass exceptions derived from LoseAllSanityAndReasonException

You could even make an ImNotSureWhichSpecifcSanityOrReasonIHaveLostButImThisSureIAmConfusedAboutTheFollowing(*args, **kwargs) exception to throw from within the exception handler

Which calls into an ordinary external Python/C++ library that parses stdout/stderr back into the machine learning JIT to obtain an estimate of how confused it is about what, specifically

If you need more data than every compiler of every language running in parallel, you can ask the NSA for its back and real time archives of every Veeky Forums post and its associated IPs and post ID / timestamp cross references

Or you could peer into your local hedge fund's Veeky Forums-scraping market data venue

You could also make a IAmNotConfusedAboutAnythingButIAmSadBecauseILostOnPerformanceAndOrAccuracyToTheFollowingOtherWorkUnits(*args, **kwargs) exception, or some multiclass combination of the two. The ML also could even generate the exception hierarchy itself

You could even parametrize the sadness by flavors and intensities and have the exception write an English poem about it on demand, so long as you don't confuse this with actual human emotion

If you really squint, a machine learning model trained on the output of these exceptions as data could probably learn to write an English poem about why it's sad about any given English language subject

Well, I'M fucking interested.

Expatiate.

You could even create a UI for this for sad humans to intellisense style write a poem about their own particular existential despair

Please and thank you.

You are a treasure, and a pleasure.

I call it, Microsoft Poem. I was going to name it Microsoft Sentence after Microsoft Word, but that has some bad implications to it.

This paper actually reminds me of my favorite song: youtube.com/watch?v=br517ctCUCE

Obviously you can also just change the sentiment flag from sad to happy to change the genre of the poem, but don't stop there. Make sure to also include the "pasta" and "zucchini" sentiments in addition to just happy or sad

If you hook this up to Google or OpenAI's cloud services, we could watch in real time our 50000th technological singularity since 2014

The only time parameter left in that case is global cloud and secret military compute capacity, which can be backed out from US equities and futures market data over the past several years

BR ratio = 1

Welcome to Humanity.

If you can't prove Yudkowsky's "Lob's Theorem" or whatever wrong using normal Aristotelian logic, can you at least prove it wrong using some or all of the alternative logics they have in the foundations of math departments in academia?

At most, I bet you'll be able to at least prove that it can't be proven wrong using Aristotelian logic. I am 99% confident that Yudkowsky is retarded, but not that retarded

If it's that 1% and he really is that retarded, I swear to god I am going to personally kill this guy

Hey asshole, it's a positional-queue.

Me first. Please and thank you.

"Anything follows from a contradiction" is a true statement, but what is this "anything", and what's a "contradiction", really? This is a question for paraconsistent logic or possibility theory or any number of things I've never studied except on Wikipedia

What's "anything"? It's the opposite of "nothing", but what's an "opposite"? I'm just guessing here based on what I know from Wikipedia

An opposite could be just about anything, but what if it's nothing? I've honestly never heard of a logic like this, nor do I expect to

Actually, I've probably heard of it somewhere, but it's probably some null object in category theory with barely any axioms or theorems in it

OK let me go nuts here for a minute - what's "an" axiom anyway? If I channel Hilbert here, an integer scalar number of axioms may only be a special case anyway

Since this is Hilbert we're talking about, even the most fully (?) general case might only be a special case after all. You never know

If you use this line of reasoning to prove the existence of God, make sure you didn't accidentally only prove it in some alternate domain of the multiverse, after which God is forced to send an army of anime characters after you to psychologically and emotionally abuse you

Your JITting AIs should compete to have a worst (by some GR metric), most inane conceivable argument that still happens to be right (by some other (?) GR metric), and see if they can beat this Veeky Forums thread

There's also a Turing test in here somewhere, though I can't precisely find it. If I were a computer, this argument would be either much scarier, or much more inane, or both at the same time in some bundle of mostly orthogonal directions

I don't know about you but all my parents beat me.

Which by the way brings me back to this video, , on a mostly unrelated subject

oh yes, except that one. Leisure Suit Larry Sierra. Tango Delta Foxtrot? Woof woof.

IDontCareIfItsEvenCorrectOrSafeIJustWantToEitherInspectOrRunThisCodeException(*args, **kwargs) is probably useful too

WtfButWhateverIJustWantToBeOnThisCitationTreeSomewhere could be a subclass of that for the peer review model

If you want to get into time derivatives, WowThisIsHappeningTooFastOrTooUnexpectedly or ImBoredHurryUp, or their higher order or multidimensional analogues could also help

>IDontCareIfItsEvenCorrectOrSafeIJustWantToEitherInspectOrRunThisCodeException(*args, **kwargs)

IDontCareIfItsEvenCorrectOrSafeIJustWantToEitherInspectOrRunThisCodeException(*syntaxIdunno, **lkih8tlkng_lk_dis)

To name some other things I'm personally into at random - the emotional and/or physical state of being tumblrkin, the tragic reality of being not just one person but an entire social graph, having -3 senses instead of 5, being outside yourself without simultaneously being beside yourself, and being anywhere or nowhere on a tensor evaluation of creativity or any other feature like the IQ scale but better (or much worse). You can also take derivatives of these over time, space, spacetime, or IPv6 IP space-time

I always hated subnetting in class.

I'm just trying to honor an old love I had: uber sub ubi (or was it ubi sub ubi?)

UBI = Taxation / Insurance (over trade)

ImBoredHurryUp(*whyaaalmao, **stahpsenpai)