pigscantfly

14 ms·

91.
by pigscantfly 9y ago
If anyone is interested in learning more on this topic, Mykel Kochenderfer's "Decision Making Under Uncertainty" offers a stellar treatment of reinforcement learning from the ground up. https://mitpress.mit.edu&#x
92.
by pigscantfly 9y ago
What about all of the grad students working and doing research? I'd argue that their labor improves society more than the average startup.
93.
by pigscantfly 9y ago
This paper checks all the boxes for me. The prior approaches are elegantly tied together, the novel contributions are straightforward and intuitive, the results are compelling, and it is extremely clearly written. Kudos to the authors!
94.
by pigscantfly 10y ago
Gratuitous misquoting by the Independent here: "But she also acknowledges that some passwords were obtained using the kind of phishing methods that hackers use to dupe people out of financial information." based on the quote &quo
95.
by pigscantfly 10y ago
No, I think errata would be more appropriate; the gist of my comment was in support of the idea behind the paper, but it's disingenuous to pretend that the model works as well as described on that task. It's also disingenuous to t
96.
by pigscantfly 10y ago
You're right, and I'm surprised (and concerned) that there haven't been any retractions or errata published for the paper cited. That said, the sentence-to-vector and paragraph-to-vector models are obvious extensions of Miko
97.
by pigscantfly 10y ago
If you enjoy allegorical, non-Western sci-fi, I would highly recommend the work of Arkady and Boris Strugatsky, two prolific brothers who collaborated on a wide range of largely science-fictional material during Soviet times. Most of their
98.
by pigscantfly 10y ago
Have you ever tried the lavender scent? I can't handle most of their fragrances for daily use, either, but that one seems significantly easier on the nose and skin for some reason.
99.
by pigscantfly 10y ago
You're right; I was arguing that French and English are more highly correlated than a pair of languages chosen uniformly at random and therefore easier for MT, but I could have made that more clear. I think the difference in accuracy
100.
by pigscantfly 10y ago
The French/English language pair was the problem machine translation research started with and has seen the most work. There are also greater similarities than most language pairs between those two due to the Norman conquest in 1066.
101.
by pigscantfly 10y ago
I'm extremely surprised they didn't outsource to Mechanical Turk or leverage their user base in some way for crowdsourcing the labels.
102.
by pigscantfly 10y ago
Stockholm is not entirely free from homelessness issues; there are lots of people - largely Roma, according to relatives and my own judgement - begging on the streets and in the train and metro stations. This is not something I used to see
103.
by pigscantfly 10y ago
As a (very anecdotal) counterpoint, I subscribed to Sonic for a year in an apartment in Mountain View near Latham and Rengstorff with absolutely abysmal results. It's been two years, so the exact details are hazy, but I believe the pl
104.
by pigscantfly 10y ago
I work on a few algorithms that could be classified as RL given an open mind. Most of them learn distributions from streaming data via some kind of online EM. I know that people in the ad-serving, porn-serving, and website optimization (A
105.
by pigscantfly 10y ago
You need to have a certain amount of karma before you can downvote comments. I'm not sure what the threshold is these days.
106.
by pigscantfly 10y ago
I would recommend trying www.writelatex.com.
107.
by pigscantfly 10y ago
Computing power absolutely does matter, because it allows us to run more complicated experiments in a reasonable amount of time, which is crucial for moving research forward. Today we work on the low-hanging fruit so that tomorrow we can re
108.
by pigscantfly 10y ago
What do you mean by expensive? Smaller neural networks are computationally cheap to run forward propagation on, which is all you need to do after deployment. This model is running in real time on your phone, after all. I'm also conf
109.
by pigscantfly 10y ago
It's interesting that the Russian state relies on these multi-party procedures (security service contacts telco to cut off service to initiate attack) rather than unilateral interception, which as the author says, might not only be sim
110.
by pigscantfly 10y ago
I've only given a couple of hiring interviews so far in my short career, but I've found specific, factual questions to be vital insofar as they truthfully clarify candidates' knowledge levels in their specific areas of expert
111.
by pigscantfly 10y ago
I was expecting to read about some experiments with autoencoders here, not this tutorial. I'm not sure how the author is learning about neural nets or where they are now, but anyone at this level of knowledge who wants to know more wo
112.
by pigscantfly 10y ago
I can't speak for TensorFlow, but I've personally found Theano to be the most easily extensible among the other widespread deep/reinforcement/gradient learning libraries (Caffe, Torch, Keras, Chainer). There does seem t
113.
by pigscantfly 11y ago
The monthly 'Who's Hiring' thread from yesterday could be a good place to start, especially if you're interested in trying out a startup job. [1] https://news.ycombinator.com/item?id=11202954
114.
by pigscantfly 11y ago
Off the top of my head and a quick search online, I can't think of what you're referring to, but a search for papers on data augmentation methods might find it. It would be helpful to know what context you're referring to -
115.
by pigscantfly 11y ago
The other country is Eritrea, actually (neighbors and in a state of frozen conflict with Ethiopia but different country).
116.
by pigscantfly 11y ago
It's not really the traditional Monte Carlo search; IIRC they're using UCT weighting for MCTS. Apologies if that's what you meant, but I think to most people "traditional Monte Carlo" means something different (pr
117.
by pigscantfly 11y ago
Have you thought about automating keyword discovery or do you plan to stick with a manually-auditable dataset? http://arxiv.org/pdf/1412.2306.pdf
118.
by pigscantfly 11y ago
I think I understand what you're getting at - that network architectures have a fixed size output - but you're incorrect in saying that the final layer must be a classifier. In general, you can optimize any differentiable functio
119.
by pigscantfly 11y ago
Caffe also supports Matlab and C++, I believe.
120.
by pigscantfly 11y ago
For all its immaturity, 4chan can be surprisingly clever. A lot of the commenters on that thread were complaining about having had accounts banned here, though.

More ›