Isak Westerlund

@westis96

ML, Robotics, Data Visualisation. Student at .

Finland
isakwesterlund.com
Vrijeme pridruživanja: ožujak 2014.
6 fotografija ili videozapisa Fotografije i videozapisi

Tweetovi

Blokirali ste korisnika/cu @westis96

Jeste li sigurni da želite vidjeti te tweetove? Time nećete deblokirati korisnika/cu @westis96

  1. proslijedio/la je Tweet
    1. velj
    Odgovor korisniku/ci

    Our intern , and I already showed this, () see also appendix. From ICLR 2020, some discussion of differences would be ideal if you can squeeze it in! Looks like a lot that's complementary inc new methods for exploiting the effect.

    1 reply 1 proslijeđeni tweet 13 korisnika označava da im se sviđa
    Poništi
  2. proslijedio/la je Tweet
    1. velj

    To the left, you see a trained agent playing a level of a game. To the right, you see the same playthrough from an agent-centric perspective: cropped, translated, and rotated with the agent in the center. Which perspective is the best input for the agent?

    13 replies 32 proslijeđena tweeta 163 korisnika označavaju da im se sviđa
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  3. proslijedio/la je Tweet
    31. sij

    Very excited to share "Learning Discrete Distributions by Dequantization" () in collaboration with and , from my internship at . We explore different methods and distributions for dequantization and reach 3.06 bpd on CIFAR10.

    49 proslijeđenih tweetova 175 korisnika označava da im se sviđa
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  4. proslijedio/la je Tweet
    31. sij

    This is a nice package for making pyplot animations more intuitive: All you do is call "camera.snap()" every time you re-do the plot.

    141 proslijeđeni tweet 562 korisnika označavaju da im se sviđa
    Poništi
  5. proslijedio/la je Tweet
    31. sij

    What unreleased FSD Autopilot sees. Straight from Tesla Autopilot recruiting website.

    231 reply 4.092 proslijeđena tweeta 13.155 korisnika označava da im se sviđa
    Poništi
  6. proslijedio/la je Tweet
    30. sij

    Examining the Benefits of Capsule Neural Networks by Arjun Punjabi et al.

    23 proslijeđena tweeta 62 korisnika označavaju da im se sviđa
    Poništi
  7. proslijedio/la je Tweet
    30. sij

    Numerically solving and backpropagating through the wave equation

    1 proslijeđeni tweet 1 korisnik označava da mu se sviđa
    Poništi
  8. proslijedio/la je Tweet
    28. sij

    Fun fact: Thinc actually implements all of the practical ideas from my "Let Them Write Code" keynote! 1. Callbacks 2. Function registries 3. Entry points 4. Single-dispatch 🔮 Docs: 🖼 Slides: 📺 Video:

    Wasn't sold on 's description of function registries for config at PyCon India (seemed a bit complex!) but after using them in thinc I want them everywhere! Config is painless and it makes it easy to do the right thing. Talk here:
    15 proslijeđenih tweetova 93 korisnika označavaju da im se sviđa
    Poništi
  9. proslijedio/la je Tweet
    28. sij

    I'm happy to announce our latest work on self-supervised learning for . PASE+ is based on a multi-task approach useful for recognition. It will be presented at . paper: code: @Mila

    7 replies 82 proslijeđena tweeta 302 korisnika označavaju da im se sviđa
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  10. proslijedio/la je Tweet
    28. sij

    Emily Black presenting the very interesting idea of counterfactual fairness: testing the fairness of a black box system by changing the protected characteristic-based distribution (eg female me -> male me) and see how it makes a difference in the outputs.

    1 reply 4 proslijeđena tweeta 20 korisnika označava da im se sviđa
    Poništi
  11. proslijedio/la je Tweet
    28. sij

    Unsupervised Disentanglement of Pose, Appearance and Background from Images and Videos pdf: abs:

    9 proslijeđenih tweetova 23 korisnika označavaju da im se sviđa
    Poništi
  12. proslijedio/la je Tweet
    28. sij

    Using OpenAI Gym and PyBullet to train an open-source 3D printed robot

    63 proslijeđena tweeta 226 korisnika označava da im se sviđa
    Poništi
  13. proslijedio/la je Tweet
    28. sij

    Towards a Human-like Open-Domain Chatbot: Near-human level open-domain chatbot is achieved with a Transformer LM with 2.6B parameters. Notably, they showed there's a strong correlation between ppl and human-likeness.

    4 proslijeđena tweeta 28 korisnika označava da im se sviđa
    Poništi
  14. proslijedio/la je Tweet
    26. sij

    Quaternions and Euler angles are discontinuous and difficult for neural networks to learn. They show 3D rotations have continuous representations in 5D and 6D, which are more suitable for learning. i.e. regress two vectors and apply Graham-Schmidt (GS).

    11 replies 145 proslijeđenih tweetova 661 korisnik označava da mu se sviđa
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  15. proslijedio/la je Tweet
    27. sij

    Short but sweet paper on recurrent autoencoder architectures for speech compression. We systematically explore the space of RNN-AEs and show that the best method, dubbed FRAE, outperforms classical codecs by a large margin. Check it out!

    i
    1 reply 69 proslijeđenih tweetova 262 korisnika označavaju da im se sviđa
    Poništi
  16. proslijedio/la je Tweet
    25. sij

    Ok folks, *now* I have a WikiArt conditional model released and a repo up! Model here: Repo here: Thanks to for their repos and their work in this space!

    8 replies 22 proslijeđena tweeta 72 korisnika označavaju da im se sviđa
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  17. proslijedio/la je Tweet
    17. sij

    New paper w , "Learning as the Unsupervised Alignment of Conceptual Systems". Supervised learning tasks can be solved by purely unsupervised means by exploiting correspondences across systems (e.g., text, images, etc.). 1/5

    82 proslijeđena tweeta 183 korisnika označavaju da im se sviđa
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi
  18. proslijedio/la je Tweet
    25. sij

    Our paper titled "Teacher-Student Training for Robust Tacotron-based TTS" is accepted by for oral presentation.🤗Such a nice way to start this weekend! Paper: Speech Samples:

    15 proslijeđenih tweetova 25 korisnika označava da im se sviđa
    Poništi
  19. proslijedio/la je Tweet
    23. sij

    Q-learning is difficult to apply when the number of available actions is large. We show that a simple extension based on amortized stochastic search allows Q-learning to scale to high-dimensional discrete, continuous or hybrid action spaces:

    6 replies 283 proslijeđena tweeta 912 korisnika označava da im se sviđa
    Poništi
  20. proslijedio/la je Tweet
    23. sij

    AI-driven deep rigging?! What if you could use AI to rig and animate a 2D video of a face? 🤯 credits to tpocomp@ on insta

    2.612 proslijeđenih tweetova 12.123 korisnika označavaju da im se sviđa
    Prikaži ovu nit
    Prikaži ovu nit
    Poništi

@westis96 još nije objavio nijedan Tweet.

Čini se da učitavanje traje već neko vrijeme.

Twitter je možda preopterećen ili ima kratkotrajnih poteškoća u radu. Pokušajte ponovno ili potražite dodatne informacije u odjeljku Status Twittera.

    Možda bi vam se svidjelo i ovo:

    ·