tshadley

11 ms·

61.
by tshadley 9y ago
> Maybe building data centers nearby Now THAT is interesting! Google is tying up hydroelectric dams with their centers currently, it seems quite feasible that massive geothermal energy could be put to work the same way. Deep-learning a
62.
by tshadley 9y ago
> I must be completely out of touch, because I could not make heads or tails of nearly any word in this article. Me too! What I struggle with most is trying to grasp where human motives fit into the game structure. Each aspect of the g
63.
by tshadley 9y ago
> But if someone disagrees with the point, they pull up the study, pick out some section of it where any type of subjective judgment call was made (usually the details of the sample, because that's the most clearly subjective thing,
64.
by tshadley 9y ago
> Peoples' perceptions of various professions are the result of socialization. But interests come first. If women on average perceive technology as not fulfilling communal goals and therefore avoid the field, they must first be int
65.
by tshadley 9y ago
She's done some research since in this area, see publication list: https://miamioh.edu/cas/academics/departments/psychology/abo...
66.
by tshadley 9y ago
> Damore points to studies showing that, e.g. women are more agreeable and more people-oriented. From that, he concludes women on average are less likely to prefer programming. That's Diekman 2010: https://www.ncbi.nlm.ni
67.
by tshadley 9y ago
https://en.wikipedia.org/wiki/Predatory_open_access_publishi... It looks like the best place to verify reputable open access journals is through the Directory of Open Access Journals: https://doaj.org/o
68.
by tshadley 9y ago
A few sentences down: "Most beginners give up when they get injured because they’ve done too much, too soon. Most of the benefits from running derive from going very slowly."
69.
by tshadley 9y ago
The study seems to hint at the idea that autism spectrum at the very mild side might be "geeky": >She added that some of the genes that contribute to geekiness and academic success might also lead to a higher risk of autism whe
70.
by tshadley 9y ago
You serialized that very well.
71.
by tshadley 9y ago
Assuming Honnold has perfect ability, what are the risks of the route itself? I assume these routes are kept clean of loose rock, holds are inspected constantly and anything showing wear is communicated to the climber community.
72.
by tshadley 9y ago
> Eventually, the earth is going to warm significantly without the aid of mankind. What are we supposed to do about that? Are you talking about Milankovitch cycles? We're about 10,000 years past the warmest point of the last cycle
73.
by tshadley 9y ago
Federated learning leaves the data on the device, only communicating updates to abstract models. See https://research.googleblog.com/2017/04/federated-learning-c...
74.
by tshadley 10y ago
"The selected model reaches 1.12 bits per byte." ( https://arxiv.org/pdf/1704.01444.pdf ) For context, Claude Shannon found that humans could model English text with an entropy of 0.6 to 1.3 bits per character
75.
by tshadley 10y ago
Lot of stuff here: https://github.com/openai
76.
by tshadley 10y ago
Word by word would require adding prior knowledge of words into the system, and they're trying to "start from scratch" as much as possible.
77.
by tshadley 10y ago
It's a good reminder that email and video just don't have the bandwidth of a few million years of evolutionary-designed face-to-face social communication.
78.
by tshadley 10y ago
Me too! The speed and efficiency of typing, editing, saving, searching, how can paper compete with that? Vim just needs a graphics plugin and we're set.
79.
by tshadley 10y ago
I'm looking at the top graph (EU 28 countries) and not seeing anything alarming relative to prior years. I'm not sure why Europe should be disintegrating now and not a decade ago if the claim is solely that Europe's problems
80.
by tshadley 10y ago
It's a worthwhile question and one worth comparing to the US today. Why is a sizable percentage of the US today afraid of immigrants despite no significant increase in immigration at all (unlike Europe)? In this case, assimilation is
81.
by tshadley 10y ago
> Then why is Europe experiencing such a huge problem now and not then? It seems less likely the problem is immigration and more likely at last part of the problem is fear of immigrants. Studies such as the following show that anti-immi
82.
by tshadley 10y ago
> You can't have short-term, mass migration without having something happen or change. I am generally confused by the claim of "mass migration". As this chart shows, EU migration of 2013 is below the rate of 2002 and is
83.
by tshadley 10y ago
> Neural networks (NNs), recently referred to as deep learning, only work "effectively" with data that is produced from a process of a continuous function. I think the author is overlooking the fact that images are not continuo
84.
by tshadley 10y ago
> Thinking about math or science for a few hours energizes me, but doing the same for social issues wears me out. I think your exhaustion comes from the moral labeling always connected to social issues. Strip out the morality and social
85.
by tshadley 10y ago
17 years doesn't seem too far off. Tim Detmer's detailed analysis puts it anywhere from 2037 to 2078, with his preference the latter (or later). http://timdettmers.com/2015/07/27/brain-vs-deep-learn
86.
by tshadley 10y ago
> I know this this is why I said there are exceptions If indeed it is physical health that is the crucial factor in preventing the decline in fluid intelligence, and older workers are becoming more healthy (relative to the older workers
87.
by tshadley 10y ago
A fluid intelligence decline is not necessarily age-related, see for example "The effect of age on fluid intelligence is fully mediated by physical health." 2013, https://www.ncbi.nlm.nih.gov/pubmed/23540273
88.
by tshadley 10y ago
Here's a chart: https://en.wikipedia.org/wiki/Comparison_of_deep_learning_so... I'd recommend starting with Theano, Tensorflow, anything python. I do like Torch's raw speed on CUDA though.
89.
by tshadley 10y ago
> "this time it's different" It may be different for other reasons but the main difference I note today is the number of opensource AI/ML platforms that are trivial to install, use, play, experiment at pretty much nea
90.
by tshadley 10y ago
> By thinking about death – even the seemingly unbearable death of one’s own child – within the much wider context of a series of natural and inevitable processes of birth and decay that permeate all aspects of the cosmos, from microbes

More ›