suddensleep

I forwarded this on to someone I know who would be perfect for the DS role; on a separate note, it is _very_ cool to see the amount of effort you put into the landing page for that role. A veritable diamond in the rough of all these copy-pasted wasteland job reqs. Here's to you finding the perfect fit!

In my experience working through the first few chapters, I'll say that the attack game framework is pretty standard across lots of course materials from universities (at least the ones that I've found posted online). One thing that is not consistent is the notation used; it seems like there are multiple competing (but essentially equivalent) sets of notations used in attack game/advantage discussions.

I've worked through the entire Paar book (which is excellent), and I've made a number of attempts at getting through Boneh and Shoup (which is also excellent). I will say that there is a good overlap in content, but the Boneh/Shoup is solidly graduate-level, whereas Paar is a good and solid introduction for an undergraduate student.

Plug for a friend's book that is forthcoming later this summer, "Topology: A Categorical Approach". [1]

I can't speak to its contents per se, because there isn't a preview yet, but I can speak to the quality of exposition in the lead author's math blog. [2]

I haven't ever dug too much into category theory for its own sake (usually just one-off chapters or appendices that get included in books on other topics), but my understanding is that it unites a lot of mathematical topics. As such, this book might be of more interest to you than, say, a classical point-set topology text, given your desire to uncover connections. That being said, there may be other category-theory-flavored books on other more strictly algebraic topics that would suit your fancy more.

[1] https://mitpress.mit.edu/books/topology

[2] https://www.math3ma.com/

I stand wholly corrected!

"In December 1840, Abraham Lincoln and four other Illinois legislators jumped out of a window in a political maneuver designed to prevent a quorum on a vote that would have eliminated the Illinois State Bank."

Imagining something like this happening today is ... difficult, to say the least.

If I've researched this correctly, I believe that "Nikoli puzzles" are not a formal class of puzzles; rather the terms refers specifically to the publisher of these puzzles [1]. KenKen, on the other hand, was invented and trademarked separately by Tetsuya Miyamoto [2].

That being said, KenKen does seem to fall into a similar class of constraint-based, language-agnostic games.

[1] https://en.wikipedia.org/wiki/Nikoli_(publisher) [2] https://en.wikipedia.org/wiki/KenKen

I was at NeurIPS last year, and I can actually confirm that the line to see Flo-Rida went around the block. It's unclear to me whether or not there was genuine interest in the performance, but there was certainly interest in the event.

Say what you will about the world's biggest nerds, but irony, whether purposeful or inexplicably accidental, is not lost on them.

Ninja Edit: New conference name :)

This is great, thanks for sharing! I wish I had known about Sol when I used to teach high school geometry; the fact that many of these constructions can be interpreted pretty openly would have produced a lot of different "answers" and opened up the floor for conversation about rigor.

Instead, I had my students construct the flag of Nepal (see http://www.servat.unibe.ch/icl/np01000_.html for the specification). Let's just say that most of the students were not as captivated by the idea as I was.

Wondering the same! It looks like Part I has been around since late 2015/early 2016 based on the course reviews.

Not a cryptocurrency course per se, but Dan Boneh's course on Cryptography[1] is an excellent introduction to most of the building blocks of cryptosystems, including the technology underlying most cryptocurrencies.

In terms of level, it is more than a little technical (programming exercises in both cryptography and cryptanalysis await you!), while still remaining far from rigorous (compared to, say, a graduate-level cryptography text).

[1]https://www.coursera.org/learn/crypto

Only marginally related, but I spent one of my summers at PROMYS[1] doing research into which complex numbers could be successfully used as bases. As different as the various common integer bases "feel" in terms of hand-computation, there's nothing like cranking out conversions of fractions to base 1+i to make you realize that there's a whole wacky universe of number representations out there. I thought it was a really cool field of study, and I remember wishing there were more readily apparent applications. Anyone?

[1] https://www.promys.org/

This reminds me of my first Geocities website: a full repository of KoRn's lyrics to date.

It's not that this didn't exist elsewhere on the internet (indeed it did, as of course I used these other sites as source material), but nowhere seemed to have the exact red-text-on-black-background look I was going for at the time.

The most excruciating part of this memory is not that I worshipped a nu-metal band, but instead that I hadn't yet discovered the magic of copying and pasting text. That's right: everything, from the lyrics themselves to the HTML tags, were typed manually by yours truly into the raw HTML editor.

I shudder to think how quickly I'd be fired today if I hadn't learned how to properly use a modern keyboard.

Disclaimer: I work as a data scientist at a large financial institution based in the US; to be fair it is one that has a reputation for being on the "cutting edge of tech", FWIW.

The data teams I work with don't know SAS and some members aren't even that facile with Excel, opting to use Apache Spark (mostly Python, but also Scala bindings) or pandas instead. Almost no one does serious data work on their local machines, and there is a big push to store all of our data off-prem.

The dev teams I work with actively experiment with different cloud-based architectures, devops automation tools, database solutions, etc.

From a product perspective, lots of teams use agile workflows, but each team is allowed to (and encouraged to) choose their own style of getting work done.

This is not meant to imply that banks have largely moved to this model; I think that we are the exception rather than the rule. This is _also_ not meant to imply that top-down corporate "solutions" don't affect us, and that greed has been completely factored out of the equation. But I've been pleasantly surprised by how much leverage and freedom we have as the "tech department".

Shadow options are the only way I can play Sudoku as well; this is also potentially why I don't enjoy playing Sudoku. It always feels like the only way to make progress is by making a number of informal suppositions and waiting for a contradiction to reveal itself. This is surely true of a large class of games, but the only type of constraint you have in Sudoku is "no direct collisions" (i.e. two instances of the same number can't appear in the same row, column or subregion). Since all of these constraints are essentially of the same class, I can't keep any more than one or two of them in my head at a time, hence "shadowing". When I compare this to a game like KenKen, where there are a number of different classes of constraint (i.e. all of Sudoku's, as well as the mathematical constraints introduced by KenKen's rules), I find that I can hold a larger number of those constraints in my head at once, even when they have non-local effects.

I only know a couple strategies for Minesweeper, so it was pretty cool to see this writeup make distinctions between local and global approaches, not to mention calling out explicit situations where guessing is necessary.

My intuition says yes; if a wrong number is placed in a square and then taken as ground truth for solving the rest of the puzzle, it certainly seems like the error would propagate.

As a concrete example, say you misplace a '2' somewhere within a given puzzle. Obviously, this cell is incorrect. But depending on the nature of what the NN has learned, it may believe the row (resp. column, 3x3 box constraint) already has the '2' in it, so tries to fill its correct spot with another number. Which of course then leads to the column and/or 3x3 box of that cell to learn an incorrect value, starting the process over again.

This same phenomenon can be seen in the game Kenken; depending on the strategies you use at any given point in the game, one mistake can propagate outward pretty quickly and spoil large sections of the puzzle.

I'll speak up from the (seemingly underrepresented) data science side here, despite not making the 3+ years ago cutoff.

I attended a data science bootcamp almost two years ago now. It was 12 weeks long and ran 9-5 each day with a mixture of lectures and pair programming exercises in the morning, and time set aside in the afternoon for working on projects. My cohort was very diverse; there were kids just out of grad school, teachers, actuaries, data/business analysts, and even practicing software developers all taking the same course. I came into the bootcamp with a fair amount of background knowledge (Bachelors/Masters degrees in Math as well as a ~10 year history of teaching myself various computer science concepts and languages), and I have to say that this served me quite well. I didn't struggle to learn Python (the language of choice for this program) or grapple with what gradient descent was really doing, because these were already parts of the way that I understood the field. Instead, I used my 12 weeks to learn about Git/Github, get really good at actually working from the command line, learn about different "big data" techniques and database structures, and pursue a passion project.

That being said, throughout the bootcamp I was keenly aware of the fact that no one was going to "fail out". There were students that needed more direct guidance than others when difficult topics were broached, and there were students whose presentations revealed that their project hadn't worked as well as they'd hoped (this includes some of my own projects). On the one hand, it was good to have a community of people (students and instructors alike) who embraced these failures and helped you learn from them. On the other hand, it instilled some level of self-doubt: "Maybe I am wasting a solid amount of my life savings on an experience that will only teach me how bad I am at this." Or even, "I feel like I did well with this project, and I have some validation from my peers and mentors, but what would a future boss think of this work?"

As a practicing data scientist now, I feel like the bootcamp prepared me to both know how to ask the types of questions data scientists ask, and to know where to look for the answers I need. As far as I know, everyone in my cohort is employed as a data scientist now, save a couple individuals with visa issues (and these few are still actively working on personal projects). Those with prior exposure to the field were certainly able to get better jobs, and quicker.

Ninja edit: italics

From the "Ethics" section of the Lyrebird site:

"Voice recordings are currently considered as strong pieces of evidence in our societies and in particular in jurisdictions of many countries. Our technology questions the validity of such evidence as it allows to easily manipulate audio recordings. This could potentially have dangerous consequences such as misleading diplomats, fraud and more generally any other problem caused by stealing the identity of someone else.

By releasing our technology publicly and making it available to anyone, we want to ensure that there will be no such risks. We hope that everyone will soon be aware that such technology exists and that copying the voice of someone else is possible. More generally, we want to raise attention about the lack of evidence that audio recordings may represent in the near future."

I'm glad the authors addressed this issue pretty forthrightly, but part of me wishes they'd written a bit more about exactly your point. Whether or not recorded speech will continue to be legally binding evidence, I think it's just as important to point out that many people are normally quite happy to take what they hear as solid evidence, especially when it aligns with their prejudices.