alexcnwy

20 ms·

1. 211.▲
   by alexcnwy 8y ago

   thanks for the tip!
2. 212.▲
   by alexcnwy 8y ago

   CNNs are usually much faster than LSTMs.
3. 213.▲
   by alexcnwy 8y ago

   I legit registered aistockphotos.com when I first saw that paper.
4. 214.▲
   by alexcnwy 8y ago

   you could use a pretrained object detection model with some lightweight logic built around it (e.g. send you a slack notification when a human object is detected in at least 5 consecutive frames between these hours) but you might also want
5. 215.▲
   by alexcnwy 8y ago

   Totally agree - it's hard to say which sequence architecture works best for your problem (are there long term dependencies, do they exist at multiple levels of scale, etc) and dataset size. Convolutions are cheap to compute and can be
6. 216.▲
   by alexcnwy 8y ago

   Shakir is a force of nature. He is one of the organizers of the Deep Learning Indaba - an incredible effort to strengthen machine learning in Africa. Unfortunately the 2018 videos aren't out yet (Shakir gave an amazing talk on deep lea
7. 217.▲
   by alexcnwy 8y ago

   Colab is awesome but I've found that GPU memory allocation can be a bit unreliable at times. It's advertised as providing you a GPU but in reality I think you're sharing the GPU with other machines which can lead to out-of-me
8. 218.▲
   by alexcnwy 8y ago

   You can use the model's output probabilities to optimize QA routing - any cases with less than 90% confidence to humans but save humans doing the cases that can easily be done by models :)
9. 219.▲
   by alexcnwy 8y ago

   Neural net universe is in Python and you can use Python to build production pipelines.
10. 220.▲
    by alexcnwy 8y ago

    Very very cool! The thinking behind improving the data preparation phase to avoid a mess of scripts that reinvent the wheel is so on-point, especially for satellite data where preprocessing can get messy. I'm looking forward to trying
11. 221.▲
    by alexcnwy 8y ago

    Sounds awful! I've been working my butt off with some friends on a system to extract data out of scanned forms and documents (incl. handwriting) and I'd really appreciate hearing your thoughts about it. Couldn't find your con
12. 222.▲
    by alexcnwy 8y ago

    Also remember businesses are taking risk - they may fail in making a profit but your income is (barring them going bankrupt) risk-free. Profit is the risk premium.
13. 223.▲
    Introduction to Machine Learning for Coders(course.fast.ai)

    2 points by alexcnwy 8y ago|0 comments
14. 224.▲
    by alexcnwy 8y ago

    Same here! Still busy with my MSc thesis though because I'm too busy applying some of the other stuff I learned from Fast.ai on other projects ':)
15. 225.▲
    by alexcnwy 8y ago

    This is incredible! I wonder if seeing yourself dance like this might speed up learning to actually dance like this...
16. 226.▲
    by alexcnwy 8y ago

    Very cool! I'm currently busy with my MSc thesis on learning to link plain text documents to semantically relevant Wikipedia articles (and some of the cool machine learning-y things you can do from there). I have 2 questions about your
17. 227.▲
    by alexcnwy 8y ago

    Any word on pricing?
18. 228.▲
    by alexcnwy 8y ago

    Casually give an informal version of the talk to some friends or colleagues to see if they "get it". I find that talking it through a few times with someone will help you figure out which parts of your explanation are unclear and
19. 229.▲
    by alexcnwy 8y ago

    Very cool! Does anyone know any good global free satellite imagery datasets?
20. 230.▲
    by alexcnwy 8y ago

    This is highly significant because it was one of the games that the Atari games that the original Deep Q learning model could not beat.
21. 231.▲
    by alexcnwy 8y ago

    I've been wanting to do an analysis of lifespans of Nobel vs the rest of the population since they seem to live longer than average. Of course, you have to be alive to win so there's a survivorship bias but it'd be interestin
22. 232.▲
    by alexcnwy 8y ago

    I think a CNN can handle scale invariance e.g. With max pooling
23. 233.▲
    by alexcnwy 8y ago

    Dresden and Budapest much better options nearby :)
24. 234.▲
    by alexcnwy 8y ago

    Bartow & Sutton is excellent. You can definitely find it online but be sure to find the right version - the latest version has great illustrations and is a lot clearer. Also, check out the RL jupyter notebook here by my friend Ryan Swek
25. 235.▲
    by alexcnwy 8y ago

    I think you totally miss the point. Explaining something by analogy can be the most effective way of "getting it". Everyone is different. You definitely cannot make such a bold statement about whether the approach was valid.
26. 236.▲
    by alexcnwy 9y ago

    I don't think you understand what dimensionality reduction means. SSN is a lookup key into the raw data. Dimensionality reduction is by definition lossy since it's used in scenarios where: rows of data = n <<< m = number
27. 237.▲
    by alexcnwy 9y ago

    You're throwing out buzzwords instead of addressing the response. It's dimensionality reduction. You cannot recover the original object. It's like using a shadow to reconstruct the face of the person casting the shadow. Note
28. 238.▲
    by alexcnwy 9y ago

    I have to disagree - it could be the case that the algorithms are sufficient and all we need is more data and more compute. I'd strongly recommend checking out Peter Norvig's talk on "The Unreasonable Effectiveness of Big Dat
29. 239.▲
    by alexcnwy 9y ago

    This great interactive really helped me understand what a convolution is: http://setosa.io/ev/image-kernels/ Highly recommended!
30. 240.▲
    by alexcnwy 9y ago

    I think about it as a 2-D weighted average. Not certain that's the correct intuition but it makes sense to me haha.