Oxford Applied AI Lab

@a2i_oxford

We explore core challenges in AI and Machine Learning to enable robots to robustly and effectively operate in complex, real-world environments.

Vrijeme pridruživanja: prosinac 2018.

Tweetovi

Blokirali ste korisnika/cu @a2i_oxford

Jeste li sigurni da želite vidjeti te tweetove? Time nećete deblokirati korisnika/cu @a2i_oxford

  1. proslijedio/la je Tweet
    4. velj

    camera-ready version of GENESIS is now online: We would like to sincerely thank the reviewers for their thoughtful comments and useful suggestions :)

    Poništi
  2. proslijedio/la je Tweet
    30. sij

    Interested in keypoint learning in radar? Excited to share some recent work with on keypoint learning for odometry estimation and metric localisation using a !

    Poništi
  3. proslijedio/la je Tweet
    28. stu 2019.

    Can we train agents that are robust to unseen distractors? Together with , we introduce APRiL, combining attention and privileged information to efficiently learn policies that generalise.

    Prikaži ovu nit
    Poništi
  4. proslijedio/la je Tweet
    18. lis 2019.
    Prikaži ovu nit
    Poništi
  5. proslijedio/la je Tweet
    18. lis 2019.

    now has a Github page showing all our open-source releases:

    Poništi
  6. 1. lis 2019.
    Poništi
  7. proslijedio/la je Tweet
    30. ruj 2019.

    I’m looking forward to discussing the impact of machine learning on the legal ecosystem with and . Are a thing to be aware of? 🤔

    Poništi
  8. proslijedio/la je Tweet
    1. lis 2019.

    We updated the GENESIS paper on arXiv, featuring more extensive experimental evaluation as well as some other changes This is joint work with , , and done at

    Prikaži ovu nit
    Poništi
  9. proslijedio/la je Tweet
    11. ruj 2019.

    If you care about artefact-free radar embeddings and state-of-the-art radar odometry take a look at recent work by and Rob Weston.

    Poništi
  10. 4. ruj 2019.
    Poništi
  11. proslijedio/la je Tweet
    4. ruj 2019.
    Poništi
  12. proslijedio/la je Tweet

    Our new Panda arm from is ready for research! We're excited to see what the Soft Robotics Lab and come up with!

    Poništi
  13. proslijedio/la je Tweet
    6. kol 2019.

    MOHART is an end-to-end multi-object tracking framework using self-attention to capture interactions between individual objects. A pleasure to be working with , , Li Sun and :

    Poništi
  14. proslijedio/la je Tweet
    31. srp 2019.

    What if we could learn to decompose and generate 3D scenes while accounting for relationships between individual objects in an unsupervised way? Turns out we can: , Oiwi Parker Jones

    Poništi
  15. proslijedio/la je Tweet
    31. srp 2019.

    Our new work on GENESIS introduces an object-centric, generative latent-variable model that performs both unsupervised decomposition and generation of visual scenes by explicitly modelling the relationships between scene components:

    , , i još njih 2
    Prikaži ovu nit
    Poništi
  16. proslijedio/la je Tweet
    31. srp 2019.

    We developed MOHART-an RNN with attention mechanisms for multi-object tracking, where several trackers communicate with each other using self-attention. Great job , et. al.:

    Poništi
  17. proslijedio/la je Tweet
    31. srp 2019.
    Poništi
  18. proslijedio/la je Tweet
    16. srp 2019.

    Robots Thinking Fast and Slow - a digest of my 2018, and 2019 talks can now be found here...

    Poništi
  19. 8. lip 2019.

    If you are in Paris tomorrow, why not drop by the Workshop on Unsupervised Learning for Automated Driving? 's Rob Weston will talk about self-supervision and inverse sensor modelling...

    Poništi
  20. proslijedio/la je Tweet
    3. lip 2019.

    Come join me at on 10 June where I will be speaking about one of my favourite topics: Robots Thinking Fast and Slow.

    Poništi

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:

    ·