angusb

10 ms·

31.
by angusb 10y ago
Ah interesting! Blind spot in my knowledge right there, thanks for pointing it out
32.
by angusb 10y ago
Can't really talk about features on here :(
33.
by angusb 10y ago
Sadly not. I'd be totally up for open sourcing if there's clear demand. If you can find it, send me an email at angus@{company_I_work_at}.com Note that it's very tied down to our use case right now: only compatible with Logis
34.
by angusb 10y ago
Can't really talk about features on here. Any smart fraudster should be watching every single thing I say :) We're using logistic regression not because it performs the best, but because it's the most understandable. When cas
35.
by angusb 10y ago
For Logistic Regression we find human readable config makes a lot of sense. It's pretty intuitive if there aren't too many features - if the model starts behaving weirdly, we can sometimes track it down to a change in a single fea
36.
by angusb 10y ago
Yep, we made our own. I haven't heard of PMML before - quite cool! What we've made is a bit more readable for what we're using it for though, IMO. Looks like this: { "intercept": 1.0, &quo
37.
by angusb 10y ago
Fraud detection at GoCardless (YC 11). We use the same tech for training and production classification: logistic regression (sklearn). ------- Training/retraining - train on an ad-hoc basis, every few months right now moving to m
38.
by angusb 10y ago
Can anyone explain why it's the ad buyers that lose out in this case, not the ad network? Surely it makes way more sense for the ad networks to bare the financial responsibility of preventing ad fraud and not the ad buyer? (The ad equi
39.
by angusb 10y ago
Are you saying that just based on the fact that they've been developing it for less time than Google, or is there a more in depth case for that somewhere? I would be very interested to read about that if there is.
40.
by angusb 10y ago
It's not just about Tesla though. They have to defend autopilot not only to protect the brand but to protect the public's perception of autonomous vehicles in general. Self driving tech is poised to save many many lives. So from a
41.
by angusb 10y ago
But that's exactly what it should say!
42.
by angusb 11y ago
Yes. New light vehicles average fuel consumption 2008: US: 11 l/100km, UK: 7 l/100km UK is pretty representative of Europe. (Also the UK improved by 20% since the 2008 figures). US and Canada are unusual in having such widespread
43.
by angusb 11y ago
This really doesn't seem like a lot. 1m cubic feet is a 30x30x30m cube. In one day it processes 24 of those. Even if there was no such thing as wind or diffusion and you only needed to treat the 30m of air next to the ground, this &quo
44.
by angusb 11y ago
Yeah, I agree. I guess their argument is that it's difficult to catch every single fraud attempt, and in this case the behaviour was just not picked up by the processor's inbuilt fraud detection systems. Still, it should be the pr
45.
by angusb 11y ago
There's also the capital cost of installing superchargers to take into account (which only supports your argument further).
46.
by angusb 11y ago
That there was a continuous stream of small bubbles (as opposed to one gigantic one) suggests that the failure mode was a leak, not an explosion. Still, I would have been scared.
47.
by angusb 12y ago
Here's a very readable quantitative discussion about tidal power generation in the UK: http://www.withouthotair.com/c14/page_81.shtml
48.
by angusb 12y ago
I know it switches from live timetable to "every 10 minutes" style estimates when: - day bus service ends and night buses begin - your phone signal dies Does it do it at other times too?
49.
by angusb 12y ago
Did anyone else find "Musicians you love listen to what you like" pretty difficult to digest? There's probably a more user friendly way to say that. Cool idea though.
50.
by angusb 12y ago
Replying because I don't think the explanations you've got so far are easy enough to read || accurate. Here's my understanding: Supercritical steam is a special form of steam that can not be described as a gas or a liquid. It
51.
by angusb 12y ago
I know that there's a kick to be got out of circumventing the draconian rules big music/film industry lobby into law, but what do the writers think about independents that they effectively take down in the same blow? This is a gen
52.
by angusb 12y ago
Ah cool, so not a problem. Out of interest how long does the camera-moving step take?
53.
by angusb 12y ago
One limitation of this that nobody has mentioned yet is that if you have to pan your cameraphone as you are taking the photos to generate the depth map you will have a harder time composing your photo than you would with a traditional photo
54.
by angusb 12y ago
A couple of other really cool depth-map implementations: 1) The Seene app (iOS app store, free), which creates a depth map and a pseudo-3d model of an environment from a "sweep" of images similar to the image acquisition in the ar
55.
by angusb 13y ago
I just spent a while looking up the maths on this and was really surprised to find that the terminal velocity of a spherical rock (diameter 10cm, density 2.5g/cm^3) at this altitude is remarkably close to 300 km/h (I got 340). Asi
56.
by angusb 13y ago
Sadly you're right. But my point grapples with a larger issue too: should a startup launch if they aren't confident about the safety of their product?
57.
by angusb 13y ago
I liked this article, and this is only a small point in the context of how interesting the rest of it is, but it would be reckless to say that you need to exceed only 1M driving hours accident free to be better than humans. Of course this w
58.
by angusb 13y ago
I think the idea was that instead of simultaneously charging several cars, it charges them one by one (round robin). You just have multiple connectors to stop employees having to leave the office to change the connectors over when one car f
59.
by angusb 13y ago
Useful point of reference: the VW golf (Europe's most popular car) sells around 500k/yr, so the Leaf's sales are about 7% of that. Seems significant.