mdda

9 ms·

31.
by mdda 8y ago
Perhaps people are wondering how your UNet is going to see the bristles on the other side of the dandelion seed.
32.
by mdda 9y ago
Look at it from an interview perspective. If I ask "are you interested in exploring ML", and you're enthusiastic, my next questions are : What have you done? Have you taken any courses? GitHub? Blog Posts? If the answer i
33.
by mdda 9y ago
Hmm. There are weekly similar counterexamples among lottery-ticket purchasers. However, don't invest your life savings in the lottery - the optimism factor is a killer. Better to invest your savings / future in something more r
34.
by mdda 9y ago
I saw the same thing being done with a theatre in Manhattan in 1998 (to make way for the Disneyfication of Times Square) : https://en.wikipedia.org/wiki/Empire_Theatre_(42nd_Street)
35.
by mdda 9y ago
I don't know of any chat logs. But it was very interesting to see the 'next response' candidates for the two sample chats in Table 1 (p3 of the PDF). In particular : it was alarming to see how much their Deep Learning respon
36.
by mdda 9y ago
And there are even a couple of photos within the closed bit of the Station : https://www.google.co.uk/maps/@51.5046516,-0.1479124,3a,90y,...
37.
by mdda 9y ago
He may have sent an actor along to represent him for the interview / photoshoot.
38.
Why Nvidia Is Building Its Own TPU(forbes.com)
1 points by mdda 9y ago|0 comments
39.
by mdda 9y ago
At a typical 'Ballroom beginners class', there will be no-one there who can already dance (i.e. if they were already great then they'd be in the next class up, etc). But also most people will have two left feet : Happy klutz
40.
by mdda 9y ago
Do you have to pay tax on your German earnings (earned in Germany, which you already pay tax to the German authorities on)? Almost all countries only tax you on the money earned (approximately) within their borders. So for most 'expa
41.
by mdda 10y ago
> NN for image classification task You've created your own straw man here. > "You realize now how computationally intractable this task is on modern hardware?" Here are the people that prove it isn't computationall
42.
by mdda 10y ago
People have known that training NNs (for any purpose) using evolution works well since the 1990s. The rise of the NN frameworks has made doing differentiation much easier now than it was before (and having gradient hints is intuitively a g
43.
by mdda 10y ago
But the judge wasn't using the newspaper report as part of his consideration. He was just offering it up as something that the defendant should be willing put into the court record (instead of taking the Fifth). The fact that it appe
44.
by mdda 10y ago
Home (lead is a teenage girl, other major character is not a love interest, nor actually male). Up (about a guy escaping being sent to an old people's home .. or something).
45.
AMD Radeon Instinct Tech Briefing(techarp.com)
1 points by mdda 10y ago|0 comments
46.
by mdda 10y ago
I can't speak for llimllib, but I didn't read their comment as being a reward for having a bath. The key bit of the suggestion is that having a bath is implicit in the whole situation. Similarly, me (+spouse) have never offered
47.
by mdda 10y ago
Wow - this brings back memories... When I was an undergrad, I took a summer job at Perihelion writing demos for the ATW. From what I recall, things weren't going great, and most of the staff was grumbling about the Atari side of the m
48.
by mdda 10y ago
Rather that talk to other people, you're literally talking at them.
49.
by mdda 10y ago
It's just after 5m45s : https://youtu.be/yq0ecBGg5Q0?t=5m45s
50.
by mdda 10y ago
FWIW, my guess is that a lot of the novel stuff that has been released in the last week is because of the impending ICLR deadline (Friday). The review process for that conference allows the papers to be updated until the reviewers' de
51.
by mdda 10y ago
I'm originally from the UK, but feel that 'healthy skepticism' there can all too often be the label that people give "why bother, it'll probably fail anyway". The boundless optimism of the US may be more benef
52.
by mdda 10y ago
I gave a talk a PyConSG this year[1], which included a demonstration of training a Reinforcement Learning model on a 'Bubble Breaker' game. There's also more detail available[2]. The Jupyter notebook is included in the GitHu
53.
by mdda 10y ago
Are they different by a factor of 10 or more? If not, then they're within an order of magnitude (base-10).
54.
by mdda 10y ago
It depends on whether the simulations are on fixed-length 'books' (the analogy used in the post) or on something more self-structuring (like Genetic Programming as a simple model, or the actual case of DNA coding for protein &#x2F
55.
by mdda 10y ago
You'd also need to robustify the compiler itself to bit-wise errors, and its output.
56.
by mdda 10y ago
But also imagine that the algorithms you write had to work (perhaps slightly differently, but essentially the same) when hit with bit-flips in the code. How would one do that? Perhaps have multiple copies, or coding redundancy, or fix-up
57.
by mdda 10y ago
The author (and everyone) should play with Genetic Programming. This is where programs (i.e. expression trees) get to mix together to produce new expression trees, according to their 'fitness'. It's an extension of the regu
58.
by mdda 10y ago
If you put your discrimination detector on an API, you would enable the original model's creators to eliminate bias by training against it. Resulting in an anti-discriminative model, through a generative / discriminative process.
59.
by mdda 10y ago
s/TensorFlow/a deep RNN like this/ would make more sense. TensorFlow is just a framework (as are Theano, Torch or DL4J) for expressing the network architecture. Framework:Network ~ ProgrammingLanguage:Algorithm
60.
by mdda 10y ago
But if you added (say) a spring into the linkages, the rest-position could be tuned mechanically to be more energy efficient (e.g: zero power). The effect of the spring could be backed out in software. But it's not until people have p

More ›