psandersen

13 ms·

61.
by psandersen 8y ago
Thanks for sharing, I can definitely relate. Good to know others have the same experience. I've also been trying to think how this can be used to design AGI, the goal being a machine that is able to recursively reason about it reasonin
62.
by psandersen 8y ago
Light quality has been something I've been interested in for a while so I've invested in making my whole environment use LIFX and Yeelight bulbs. This has been great and I get bright neutral lighting during the day, warmer light i
63.
by psandersen 8y ago
I'd like to see a serious investigation and discussion around the recommendation engines used to rank content added here. Something is seriously wrong with youtube's algorithm, whether in the technical implementation or how it dea
64.
by psandersen 8y ago
You mean average skill has increased?
65.
by psandersen 8y ago
This is a pretty essential point that I think changes the calculus of contact quite a bit. A civilisation capable of visiting us at a reasonable expense to themselves would presumably have much easier access to near unlimited astroids/
66.
by psandersen 8y ago
Absolutely, its a novel and there are around 50 trillion 'baseline' humans not that different from us. Banks himself has said he has focused on writing more about people who are edge cases or less normal in the culture to keep thi
67.
by psandersen 8y ago
Data science is pretty fun and quite well paid. I enjoy my work a lot and manage a good work life balance.
68.
by psandersen 8y ago
This is a great comment, and I just wanted to add that imagining what a fictional high tech utopia would look like was what inspired Iain M. Banks to write the culture series. For anyone interested in tech and the far future its a must read
69.
by psandersen 8y ago
The reaction speed should be random (80-200ms or more - maybe quite a tail latency) instead of fixed during training to be more consistent with human behaviour and allow abilities to sometimes work so the bots are more likely to sometimes t
70.
by psandersen 8y ago
So much this, even if you can create a decently accurate ML system to score student essays it absolutely doesn't handle the adversarial nature of scoring and is frankly insulting given how much students are paying for tuition these day
71.
by psandersen 8y ago
Its also possible that if you have good data on a person's financial situation and behaviour (and more questionable that of their social circle) that race stops being a relevant signal in the algorithm as it was only ever a proxy for a
72.
by psandersen 8y ago
Given a pixel isn't a perfect colour dot, it has subpixel patterns and there is space between pixels; there is an advantage to having substantially higher pixel count than we can perceive. What we really need are better up-scalers and
73.
by psandersen 8y ago
I think you have to consider downside risk, if the economy goes bad or you lose your job how easily can you make rent and survive? At 75% you have much less margin than at 22%.
74.
by psandersen 8y ago
Agree completely with this, tidyverse is whats keeping me in R when I'd prefer to mostly use Python.
75.
by psandersen 8y ago
Which is frustrating, I wouldn't blink at paying that much for Facebook if it meant a genuine ad and tracking free experience. I want to be your customer, not your product. IMHO Europe's GDPR should be mandating that companies are
76.
by psandersen 8y ago
Awesome, thanks for the reply. So if someone had a cockroach or ant connectome the Nengo simulator could, in theory, be used to simulate a population of them connected to some virtual environment?
77.
by psandersen 8y ago
That sounds pretty interesting, would you happen to know if these can be used to simulate the worm connectome (OpenWorm?) linked up in a virtual environment?
78.
by psandersen 8y ago
Thanks for this summary! Pure speculation, but one possible benefit of dramatically reducing the on-chip power requirements would be allowing larger areas of the chip to be active for longer periods of time. Most of the area of a cpu is ina
79.
by psandersen 8y ago
This looks really interesting! Wonder if it could be integrated into the whole OS/kernel, and if it can help with typical ML workloads like running a randomforest in scikit learn.
80.
by psandersen 8y ago
Wait, so if I've paid the subscription fee for 2 years I get the one at the start of the subscription period? Some lag from the latest version seems fair, e.g. 6 months before the latest major version is perpetual...
81.
by psandersen 8y ago
There are plenty of great answers here already, but what I think would help improve the state of the art would be configurable reward sparsity and configurable priors. - Reward sparsity means that an easier version of the game gives a score
82.
by psandersen 8y ago
Awesome, thanks for that. I'll definitely give it a shot once rstudio is supported.
83.
by psandersen 8y ago
Thanks for that, link seems down unfortunately. I use data.table a lot but would happily switch if the dplyr family of packages work. Would love to use this in Rstudio for exploratory analysis which would also need ggplot to work. Quite oft
84.
by psandersen 8y ago
This looks really interesting. Can anyone comment on how mature the R and python support is? I'm coming at this from an ML/data science background, so anything that speeds up the above while maintaining broad library support would
85.
by psandersen 9y ago
The subscription fee should be equivalent to the loss to the business from no longer collecting the data on the user. It could actually be an eye-opener for many people. Most middle-class tech people would probably happily pay $10+ a month
86.
by psandersen 9y ago
Worth adding that if we get more hardware support for deep neural networks on the GPU (e.g. tensor cores), training a denoising DNN on the noisy renders can save a lot of computation and fits neatly into the pipeline. If the ray tracing can
87.
by psandersen 9y ago
Additionally, I'd love to see a decoder add noise proportional to the uncertainty and surrounding structure as over smoothed images look unnatural as well. See activity masking demo https://people.xiph.org/~jm/daal
88.
by psandersen 9y ago
A computer still deals with a finite amount of bits (e.g. 32-bit floating point numbers) so its a discrete system that is large enough to approximate a continuous system. It may not matter much in practice except for numerical precision iss
89.
by psandersen 9y ago
Just randomize the content they're trying the scrape imho
90.
by psandersen 9y ago
I'd love to see it as an app with a blacklist for spam numbers or telemarketers. Use a better text generation engine like a modern deep learning based generative model, and deep voice or crowd-sourced recordings.

More ›