fnbr

34 ms·

331.
by fnbr 10y ago
My uncle's like that. He has no idea in exchanging ideas, he just wants to tell you what he thinks and have you agree. It's exhausting to engage with him. When he first met my fiancee, he spent an hour telling her (a clinical gene
332.
by fnbr 10y ago
I recently built a machine learning web app. When I was building it, I didn't find many resources for how to design it- I wanted to do a write up to get feedback on the code layout, and to help others in the same position.
333.
How we designed our machine learning application(finbarr.ca)
2 points by fnbr 10y ago|1 comments
334.
by fnbr 10y ago
We're technically using it in production for my site, www.bugdedupe.com, but we're still in beta, so we haven't experienced heavy usage. We've been using Kubernetes to deploy Docker containers on Google Container Engine,
335.
by fnbr 10y ago
What type of architecture are you using? We're starting to set up a distributed backend at http://www.bugdedupe.com , and there don't seem to be a lot of resources out there about how to do it right.
336.
by fnbr 10y ago
I was able to get a different model working, and FloydHub is great- I'm really finding it useful. I'm looking forward to see where you take it!
337.
by fnbr 10y ago
This seems pretty awesome, but I can't get it working with my code. Do you have any plans to release extra documentation, or examples of how to adjust existing models to run on FloydHub?
338.
by fnbr 10y ago
If you can upload a sample log, I'd be happy to take a look and try to provide some more specific guidance (email's in profile). I did some work using stack traces to predict duplicate bug reports, so I'm somewhat familiar wi
339.
by fnbr 10y ago
What you could do is assemble the data in tabular form so that your data is in the shape: Issue System log -------- ------------ issue_1 corresponding system log issue_2 corresponding system log issue_3 corre
340.
by fnbr 10y ago
With current interest rates, it wouldn't have to fall much more, but yes. Plus opportunity cost, and a return to compensate for the risk [1], etc. [1] It's very risky to short a stock as you can end up losing a very large amount
341.
by fnbr 10y ago
Yes, exactly. Ideally (for the short-sellers) the stock would fall even further, as if the stock is at the exact level it was when they bought, then they'll only break even.
342.
by fnbr 10y ago
Also, Slack- if Lync wasn't as brutal to use, Microsoft would have stopped Slack dead.
343.
by fnbr 10y ago
This would have been brilliant 10 years ago. If BBM was open to non-blackberries, it could have killed WhatsApp before it began. It's weird to think of how popular Blackberries & BBM was back in the day. I was in high school in 200
344.
by fnbr 10y ago
Sure, but doesn't that prove the point? It doesn't matter how legitimate your protest is, just that you have leverage, which the #deleteuber crowd did.
345.
by fnbr 10y ago
I'm typically a naysayer when it comes to protests, and particularly hashtag activism- I (strongly) doubt their efficacy. However, #deleteuber seems to have been remarkably effective. I wonder if this will be enough to staunch the blee
346.
by fnbr 10y ago
I get that. That was my company's experience with Slack.
347.
by fnbr 10y ago
Why doesn't this exist? It seems like an instant win.
348.
by fnbr 10y ago
What do you mean? Couldn't you sign up for a Business Class account? [1] [1] https://trello.com/pricing
349.
by fnbr 10y ago
Obviously overfitting is always a concern, but deep learning has (when done correctly) found ways to mitigate concerns about overfitting. Dropout, for instance, or k-fold cross validation.
350.
by fnbr 10y ago
I did a master's in econometrics and that was my experience. In the last year of my undergrad, I took a course in machine learning that seemed to have way more intellectual rigor than any of the material covered in grad school. At one
351.
by fnbr 10y ago
Not quite- the part of Yahoo owned by Verizon will still be called Yahoo, but the part not acquired by Verizon (essentially the Alibaba shares) will be called Altaba.
352.
by fnbr 10y ago
Really cool to see Leo doing this. If he ever moves to an online version of this, I'll be the first to sign up. Huge fan of his blog.
353.
by fnbr 10y ago
Ah, fair enough. I'd be interested in reading it, but I think that (unfortunately) you're correct that it's mostly non-replicable, at least based on my experience.
354.
by fnbr 10y ago
How do you find contracts? I'm interested in doing contract data science work, but I don't know how to start finding interested potential clients.
355.
by fnbr 10y ago
Twitter seems to be having a big shakeup this year, what with the layoffs and shutting down Vine [0]. I wonder if this is part of it. I hope it shows that Twitter is becoming more serious. They've done very little meaningful innovation
356.
by fnbr 10y ago
I'd say week rate, unless the consultants are particularly specialised.
357.
by fnbr 10y ago
Yup. That's what you'd expect to pay for a week or so for a skilled consulting firm. Not a ton of money.
358.
by fnbr 10y ago
I'd join that list!
359.
by fnbr 10y ago
I'd also be interested to hear this. I'm running a project that's 10gb in size and uploading the data to AWS S3 was absurdly slow. Is there any way to speed up the upload that you found? 10 GB was painful though, I can't
360.
by fnbr 10y ago
It's incredible to think of all of the harm that Elizabeth Holmes has created through her desire to be the next Steve Jobs. She has very probably killed people, ruined families, and caused at least one suicide [1]. And all for what? Th

More ›