alexcnwy

31 ms·

1. 301.▲
   by alexcnwy 9y ago

   Thanks for the links - a friend of mine is working on something like DeepLIFT but I hadn't heard of it...
2. 302.▲
   by alexcnwy 9y ago

   Do you have any advice on getting your own box set up in a data center - constantly traveling...
3. 303.▲
   by alexcnwy 9y ago

   Hadn't heard about compiling yourself improving performance for cloud CPU usage - thanks!
4. 304.▲
   by alexcnwy 9y ago

   I think it's in part due to the content loss piece being done using pre-trained imagenet models which typically resize images to 224x224...
5. 305.▲
   by alexcnwy 9y ago

   Offer to give talks at local meet ups on challenging subjects that only an expert would be able to talk about... and then not mess it up. Confidence will follow :)
6. 306.▲
   by alexcnwy 9y ago

   Seconded - can't recommend the course highly enough A lot of 'traditional' computer vision methods e.g. Hough detector are simply inferior to deep learning approaches. Plus, it's a lot easier than you'd think to get
7. 307.▲
   by alexcnwy 9y ago

   2 counterpoints: 1. buying data is expensive 2. running a fund is a completely different skill than building predictive models (back-office, compliance, etc.)
8. 308.▲
   by alexcnwy 9y ago

   My favourite source of useful useless knowledge: https://www.reddit.com/r/todayilearned/
9. 309.▲
   by alexcnwy 9y ago

   Yeah Central Festival was epic - not sure where you live that makes CF look like it's falling apart by comparison?
10. 310.▲
    by alexcnwy 9y ago

    I was going to suggest Chiang Mai after reading the OP then scrolled down and saw it's the top comment. I just spent 4 months traveling around Asia with 1 month in Chiang Mai. It's an incredible city and incredible in ways you wou
11. 311.▲
    by alexcnwy 9y ago

    Totally agree. Similarly for word embeddings like word2vec, GLoVE, fasttext etc in the case of NLP. I think this is fundamental - if you teach a human how to recognize street signs, you don't need to show them millions of examples - ju
12. 312.▲
    by alexcnwy 9y ago

    Can't recommend the course highly enough!
13. 313.▲
    by alexcnwy 9y ago

    I suspect you could just use the content loss approach from the style transfer examples. Take say the second last layer of a pre-trained convolutional neural network (e.g. VGG) and just use that to compute cosine similarities between images
14. 314.▲
    by alexcnwy 9y ago

    You guys are awesome!
15. 315.▲
    by alexcnwy 9y ago

    Yeah PCA will give you eigenvalues of the PCs in descending order of variance explained so just summing and weighting those tells you the first 3 PCs explain say 93% of the variance in the data.
16. 316.▲
    by alexcnwy 9y ago

    I think there's something of a catch-22 here where it's much easier to get a low-hours job when you've already put yourself out there and have something like a blog that you can point to. One thing I'd suggest is speakin
17. 317.▲
    by alexcnwy 9y ago

    Try compute the median point across all members of the cluster then compute the cosine similarity between each point in the cluster and the median of that cluster and use the most similar as your label point.
18. 318.▲
    by alexcnwy 9y ago

    Humans don't run split tests on alternate realities but they can learn from wikipedia alone - I think we just don't know quite how yet. We need a lot of data currently because algorithms don't generalize like people do. I&#x2
19. 319.▲
    by alexcnwy 9y ago

    Keras is higher level, parent asked about lower level
20. 320.▲
    by alexcnwy 9y ago

    I don't know - maybe I'm just irrationally optimistic but I don't believe anyone has "locked up all of the data". You can download the whole Wikipedia data dump at under 100GB uncompressed (text only - with media is
21. 321.▲
    by alexcnwy 9y ago

    Automatic text summarization.
22. 322.▲
    by alexcnwy 9y ago

    Ah so jealous! How was part 2?
23. 323.▲
    by alexcnwy 9y ago

    great, thanks for your reply and looking forward to that blogpost!
24. 324.▲
    by alexcnwy 9y ago

    I think this is true not just of images but also for text with things like word2vec etc
25. 325.▲
    by alexcnwy 9y ago

    For anyone keen to learn more about object detection (and deep learning in general), I just finished working through an excellent free MOOC taught by Jeremy Howard (former chief scientist at Kaggle) - you basically learn how to fine-tune a
26. 326.▲
    by alexcnwy 9y ago

    yes please - would love to read that blogpost!
27. 327.▲
    by alexcnwy 9y ago

    AWESOME! Can you please elaborate a bit more on the model architecture and what you tried with respect to transfer learning? Did you use an imagenet architecture e.g. VGG and retrain from scratch or a custom architecture? Did you try chop o
28. 328.▲
    by alexcnwy 9y ago

    Awesome! P.s. Richard Socher, one of the authors on the paper taught a great Stanford course 'CS224d: Deep Learning for Natural Language Processing' with videos and notes available here: http://cs224d.stanford.edu/