uoaei

30 ms·

541.
by uoaei 2y ago
It's also obvious and it's hacky. Frankly I'm stunned this hasn't been tried yet. The people thinking this is a stepping stone to More Intelligence are missing the forest for the trees. Deep learning is always and only e
542.
by uoaei 2y ago
I believe one example would be quantum annealers. Where "programming" involves setting the right initial conditions and allowing thermodynamics to bring you to an optimum via relaxation.
543.
by uoaei 2y ago
You're missing the point entirely: framing the swarm as "covering interstellar distances" is misleading and betrays a certain naivete in thinking about space travel that compromises the reporting on this plan.
544.
by uoaei 2y ago
ChatGPT isn't very good for knowing how well or correctly code will run
545.
by uoaei 2y ago
The solution is to destroy blind techno-utopianism, and realize tools like AI are being used in ways that are harmful to society and hard to see coming if all you think about is generating increasingly realistic-looking videos. The volume o
546.
by uoaei 2y ago
I'm not sure "swarm over interstellar distances" makes any sense here, then. How many stars are within 100,000 km of each other and what percentage of stars does that represent?
547.
by uoaei 2y ago
I appreciate linked data / the semantic web and its domain model-forward approach, labeled "ontologies". Organizing relationships between objects and actors contextualizes them and forces you to pursue parsimony. It's ni
548.
by uoaei 2y ago
Can you say relativistic corrections hell? "Over lightyears" means it takes literally years for these things to coordinate moves with each other, to say nothing of post-launch coordination messages from Earth.
549.
by uoaei 2y ago
The entire reason SGD works is because the stochastic nature of updates on minibatches is an implicit regularizer. This one perspective built the foundations for all of modern machine learning. I completely agree that the most effective reg
550.
by uoaei 2y ago
> if it was a negative return on investment, energy wise Unfortunately, cost does not track emissions that closely. People do all sorts of things that cause emissions all the time, and usually their choice is dictated by which option is
551.
by uoaei 2y ago
I agree about zoom levels on different websites when on desktop. Zoom works differently on mobile vs desktop which is its own challenge as far as ergonomic use. On mobile, to get the up/downvote buttons to the size I want them, I'
552.
by uoaei 2y ago
Accessibility isn't just about possibility, it's about ergonomics. You could integrate a differential equation by lining rocks up in a large desert as a computer, but you wouldn't say that solution is "accessible"
553.
by uoaei 2y ago
Now just imagine what we could do with a purpose-built ML model !!!
554.
by uoaei 2y ago
Because "machine" is a poor analog for government activities.
555.
by uoaei 2y ago
Indeed. An issue I ran into over and over while doing research for semiconductor manufacturing. My complaint was more illustrative than earnest.
556.
by uoaei 2y ago
"Time series" is such an over-subscribed term. What sorts of time series is this actually useful for? For instance, will it be able to predict dynamics for a machine with thousands of sensors?
557.
by uoaei 2y ago
Behavior in high dimensions is even weirder than you might think, because you're nearly always close to a saddle point. There's way more saddle points than minima or maxima (consider modeling the sign of eigenvalues for the Jacobi
558.
by uoaei 2y ago
Zulip has many of these features I believe. Anecdotally it seems to be great for running an online discussion forum for a high school or college level class.
559.
by uoaei 2y ago
"Seems to be on a tear" is the result of good marketing and well-timed releases. Obviously these kinds of things don't just appear one day.
560.
by uoaei 2y ago
Are analogies always constrained to be so literal? Do we complain about how many legs the 'wasp' has when talking about wasps and orchids?
561.
by uoaei 2y ago
> It's only boring adults who do steady state cardio That's an interesting observation, I'd never noticed that but you're right. There was always something uncanny to me about high school teams for cross-country runni
562.
by uoaei 2y ago
> The difficulty is that on average the people who sleep well and are physically active are also the people who eat well, have higher incomes and better access to healthcare Truly effective causal studies are still in their infancy, re
563.
by uoaei 2y ago
Objectification, that's how.
564.
by uoaei 2y ago
To be conscious and to be self-conscious (sentient) are different things.
565.
by uoaei 2y ago
Would two cities debating about whether humans are conscious be able to understand that we humans ask the same questions about our cells?
566.
by uoaei 2y ago
It's built on JAX, not vanilla Python. The metric being optimized is not just performance, but also the ability to build reasonably performant workflows with arbitrary differentiable (i.e., ML) inputs and outputs.
567.
by uoaei 2y ago
So you want to tax tires separately to whole cars? You don't think your scheme will induce perverse incentives as regards tread patterns, grip in different weather scenarios, etc.? If I were a tire manufacturer I'd love this becau
568.
by uoaei 2y ago
At a rate proportional to the fourth power of the weight per-axle[1]. So considering how much more EVs weigh compared to their ICE counterparts, we can only hope that compact and lightweight EVs become more prominent. [1] https:/&#x
569.
by uoaei 3y ago
Rhetoric was invented in Greece
570.
by uoaei 3y ago
This is how NN theory describes the process of training, i.e., that's the mechanism by which all neural networks come to represent the problem they are being applied to solve. GP is using a fancy way to say "train NNs on massive d

More ›