kohlmeier

4 ms·

1.
Accelerating I/O bound deep learning on shared storage(blog.riseml.com)
3 points by kohlmeier 9y ago|0 comments
2.
by kohlmeier 12y ago
I'm a bit ambivalent about that feature. I do think many folks will want/expect it, and it would be useful at times. At the same time, the primary use of the app I personally wanted is to add things I wanted to remember permanentl
3.
by kohlmeier 12y ago
Thanks for the feedback. Bulk import would be a pretty fast and simple add; I'll make a GitHub issue for it.
4.
by kohlmeier 12y ago
Not currently, though that would be quite cool. For now, I clarified my version of that card, which unfortunately had a typo. Thanks..
5.
Show HN: Memfinity.org – Social, spaced-repetition web service(memfinity.org)
23 points by kohlmeier 12y ago|9 comments
6.
Compressed Features for Machine Learning(derandomized.com)
1 points by kohlmeier 13y ago|0 comments
7.
Khan Academy: Machine Learning → Measurable Learning(derandomized.com)
17 points by kohlmeier 13y ago|0 comments
8.
The Science of Learning (a reading list)(derandomized.com)
4 points by kohlmeier 14y ago|0 comments
9.
Khan Academy user simulations: Why divide by (n-1) to estimate variance?(khanacademy.org)
7 points by kohlmeier 14y ago|0 comments
10.
by kohlmeier 15y ago
This is not naive bayes and does not assume independent observations on the exercises. The point of using a network is to model the joint distribution with dependencies.
11.
by kohlmeier 15y ago
The 'E' in the regression is the inferred/predicted value of the E variable for that exercise, using no problem history from that exercise--only what's pulled in through the Bayes net. (Sorry that wasn't clear) The 'T' variable is likely j
12.
Bayes nets by example with Python and Khan Academy data(derandomized.com)
139 points by kohlmeier 15y ago|9 comments