Related to the Coral accelerator, is it usable as a general accelerator outside AI workflows? When I looked into it, there was no general SDK on using the Coral, only an "ingest ML model" compiler.
Not surprising. It always seemed likely to me that there is model bias if you train your models on model generated data, like a feedback loop (second order effects?). Similar to how applying a linear system over and over stretches the inputs in the direction of its largest eigenvector.
Now wait till the generated content is indistinguishable from human content (to humans) and it will be hard to figure out what's in your training set.
Not a direct answer, but last year I got a 13" eink second screen and it's been the best investment I made for my eye health since getting glasses. The difference was noticeable from the start, at the end of a day without the eink screen my eyes feel tired, similar to sand in your eyes. After a day of using the ink screen my eyes feel a lot better, still relatively tired, but like the tired you get at the end of a day not in front of your computer.
Getting an eink secondary monitor for my workstation. I use to get pretty bad eye strain that lasted for days from LCD screens. It got so bad I forgot how it feels to wake up without my eyes hurting. Now it's all gone.
The product offering of eink screens is virtually non-existent, the only usable one for interactive work that I found was the Dasung 13.3", which is the one I got.
Fair point. At the same time the exact computation paradigm is proof that problems like these can be solved in a rigorous way if you start with the right foundation.
By the way thanks for writing the article, it was a refreshing read.
One geometric model that actually handles numerical approximations is called exact geometric computation. It's pretty interesting and starts by explicitly tracking arithmetic errors in the operations you do. At its core it's a mix of interval arithmetics and multi-precision floating point. This is something you can do since the floating point implementations on most (all?) CPUs give you strong guarantees on the error bound of floating point arithmetic operations and these are usually within 1 ulp (unit in the last place) of the correct result. Now you have an interval for the result and can build your geometric constructs on interval arithmetic rather than classic, single point scalar arithmetic.
Geometry works with constructs, not numbers. An intersection is a construct, collinearity is a construct, orientation / handedness is a construct, etc. If you define your fundamental constructs to be rigorous then the algorithms you build on top will also be rigorous. If we consider the article's inside-outside test example, this result would be computed based on the intersections (intervals) we've computed. Since we know for sure the correct result is within our interval of error then inside-testing boils down to comparing intervals and we know there are three possible situations here: 1. [a, b] < [c, d]; 2 [a, b] > [c, d]; 3. [a, b] overlaps with [c, d]. In case 1 and 2 we have a robust conclusion and in case 3 we recompute the whole predicate with higher precision until it becomes decidable.
The golden standard for exact geometric computation is CGAL [1]. All the research around it is very interesting and refreshing to read.
At https://www.3dhubs.com we're using Rust to build the geometry processing side of our platform, which is the most computationally intensive custom part. Previously it was in C++ and we're a lot happier after migrating to Rust two years ago.
It might also be specific to the niche you're hiring for. What's your field of operation? What technologies / technical disciplines are you looking for?
Personal website that I've been wanting to turn into a blog about computational (differential) geometry, applied mathematics, performance optimization and distributed systems.
1. La peste (Albert Camus) - entertaining if you're into French existentialism
2. Linked (Barabasi) - insightful and useful to understand that most phenomenon we see in real life follow a power law and what that means
3. Leonardo da Vinci (Walter Isaacson) - surprising how little of da Vinci's real personality is reflected in his modern image
4. Matrix computations – third edition (Gene Golub, Charles van Loan) - good read if you're into linear algebra
5. Giving effective feedback (Harvard business review) - useful for getting an understanding on how to handle human interaction
6. The old man and the sea (Ernest Hemingway) - enjoyed it a lot, a reflection of Hemingway's romantic spirit. Also, a quick read.
7. Screwjack (Hunter S. Thompson) - didn't resonate with me although I am a fan of existentialism
8. Sun and steel (Yukio Mishima) - raw and intimate. before reading this book I'd recommend reading up on Mishima's life
9. Stranger than fiction (Chuck Palahniuck) - somewhat entertaining, more so because it served as a glimpse into Palahniuck's creative process
10. Thinking in Systems: A Primer (Dana Meadows) - useful for understanding that in modern life we have complex systems at work with emergent behavior that we didn't expect. Trying to isolate / model a single component of one of these systems is a flawed approach.
11. Chaos monkeys (Antonio Garcia Martinez) - entertaining, but quite long
12. Weapons of math destruction (Cathy O'Neil) - in a way, similar to Thinking in systems, but at an applied level. Shows how rules in modern society can have unintended negative consequences when hidden negative feedback loops emerge from the complex system they are embedded in.
13. Lolita (Nabokov) - a brilliant novel from many points of view. Although the topic is controversial, it is a book that had to be written.
14. The prince (Machiavelli) - an interesting read. It was intended as a guide to the young price Cesare Borgia from his teacher Machiavelli. The secret to enjoying it is not to judge it by modern morality.
15. Give and take (Adam Grant) - psychology research about giving / taking / matching personality types presented in a book for the masses. I would not read it again; watching a presentation online should be enough to get the point across.
16. Mécanique (Landau, Lifchitz) - refresh of mechanics
17. Do androids dream of electric sheep (Philip K. Dick) - enjoyable and entertaining! I'm surprised how different the feel of the book is compared to the Blade runner films. In the book androids are purely rational beings, whereas the film wraps them in an aura of romanticism.
18. How we learn (Benedict Carey) - decades of learning research condensed in a book
19. Advanced Calculus: A Differential Forms Approach (Harold M. Edwards) - the best math book I've read so far! For me it was eye-opening in a fundamental way. Edwards is truly a gifted teacher.
20. Three men in a boat (Jerome K. Jerome) - enjoyable and amusing
21. Getting to yes (Roger Fisher, William Ury) - negotiation book, nicely written and structured
22. Without a word (Zhang Jie) - historical novel by one of China's most acclaimed modern writers. I don't have a good reference to compare against since this was my first Chinese novel, but overall I did enjoy it. It felt real and raw.
Pretty cool! How about turning this into a service where other people can lease their own GPU hardware? Either through your website or by offering it as a sort of turnkey solution.