Rezultati pretraživanja
  1. prije 3 sata

    Improving Efficiency in Large-Scale Decentralized Distributed Training by et al.

    3 proslijeđena tweeta 2 korisnika označavaju da im se sviđa
  2. 2. velj

    Can be fooled when it comes to ?

  3. 2. velj

    I Am Bless and Happy Because . first time ko nilibre si mama sa kanyang favorite na japanese restaurant sa * gamit ang aking unang 13th month pay sa aking first Job sa ❤️💜💙

    1 proslijeđeni tweet 1 korisnik označava da mu se sviđa
  4. 31. sij

    Semantic Adversarial Perturbations using Learnt Representations by Isaac Dunn et al.

    4 korisnika označavaju da im se sviđa
  5. 31. sij

    Homogeneous Vector Capsules Enable Adaptive Gradient Descent in Convolutional Neural Networks GitHub:

  6. 31. sij

    Level up your data science vocabulary: ImageNet

    1 proslijeđeni tweet 2 korisnika označavaju da im se sviđa
  7. 30. sij

    paper: "Towards fairer datasets: filtering and balancing the distribution of the people subtree in the hierarchy" going from 2832 people labels to only 159 "safe" and "imageable" labels...

    1 reply 3 korisnika označavaju da im se sviđa
  8. 29. sij

    CSNNs: Unsupervised, Backpropagation-free Convolutional Neural Networks for Representation Learning by Bonifaz Stuhr et al.

    2 proslijeđena tweeta 7 korisnika označava da im se sviđa
  9. 29. sij

    Artificial Intelligence Can Insult, Curse, Hurl Racial Slurs

  10. 28. sij

    Why is considered such a major turning point in deep learning? Our latest blog dives into our collaboration with Alectio to develop processes to improve the quality of any training set, and eventually, train better models with less data.

    1 korisnik označava da mu se sviđa
  11. 27. sij

    On Last-Layer Algorithms for Classification: Decoupling Representation from Uncertainty Estimation by Nicolas Brosse et al.

    12 proslijeđenih tweetova 24 korisnika označavaju da im se sviđa
  12. 24. sij

    Cross-Domain Few-Shot Classification via Learned Feature-Wise Transformation by Hung-Yu Tseng et al. including

    3 proslijeđena tweeta 6 korisnika označava da im se sviđa
  13. 23. sij

    "GI needs its own ". ImageNet =1.2 million images of >1000 categories, incl “tree”, “tool”, & “tractor”.. This is used as 'pre-training' for medical algorithms(!) suggests GI-specific pre-training may ⬆️ performance for endo .

    9 proslijeđenih tweetova 22 korisnika označavaju da im se sviđa
  14. 22. sij

    Towards detection and classification of microscopic foraminifera using transfer learning by Thomas Haugland Johansen et al.

    2 proslijeđena tweeta 3 korisnika označavaju da im se sviđa
  15. 20. sij

    Increasing the robustness of DNNs against image corruptions by playing the Game of Noise by Evgenia Rusak et al. including

    2 proslijeđena tweeta 8 korisnika označava da im se sviđa
  16. 20. sij

    Noisy Machines: Understanding Noisy Neural Networks and Enhancing Robustness to Analog Hardware Errors Using Distillation by Chuteng Zhou et al. including

    1 proslijeđeni tweet 4 korisnika označavaju da im se sviđa
  17. 17. sij

    I totally agree with conclusions in . We need specific datasets like for . For GI endoscopy we published just recently an open dataset that tries to be the first step towards this:

    5 korisnika označava da im se sviđa
  18. 9. sij

    Trends in Machine/Deep Learning w/ Zack Lipton 💭Premises & Hypothesis ✅Transformers instead of LSTM ⚠️Invariant Risk Minimization 🔄Generative VS Discriminative Models 🛋️Zack 🎙️Sam 📍

    2 proslijeđena tweeta 14 korisnika označava da im se sviđa
  19. 9. sij

    In 2019 a webapp called resulted in many social media posts of people tagged as criminal or gender stereotypes. It shows the importance of fighting bias in . Hence, it is good to see that gets a label update.

    3 proslijeđena tweeta 1 korisnik označava da mu se sviđa
  20. 9. sij

    The latest newsletter is just amazing. The part that I liked the most is the dataset is being modified and efforts are being made to purge the bias outta it. Read the jam-packed newsletter here: .

    54 proslijeđena tweeta 124 korisnika označavaju da im se sviđa

Nema rezultata.

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.