Jevgenij Gamper

@brutforcimag

PhD student , Warwick Machine Learning Group, Tissue Image Analytics lab. Senior statistical scientist at . Views my own.

London, UK
Vrijeme pridruživanja: veljača 2017.

Tweetovi

Blokirali ste korisnika/cu @brutforcimag

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

  1. 31. sij

    Firms in healthcare are developing AI tools using data that is mostly sourced from exlusive partnership hospitals, and the results above demonstrate there is a strong potential for biases and fairness issues. Are there any standarts to regulate it ?

    Prikaži ovu nit
    Poništi
  2. 31. sij

    At ML4Health , showed that SOTA Cest X-ray classifiers are biased to sex, age, race and insurance type. Would this hold in pathology too ?

    Prikaži ovu nit
    Poništi
  3. 31. sij

    They show that firms are incentivized to invest less in minority groups, as these are not profitable. And the competition does not mitigate this incentive towards inequality, and that REGULATION CAN IMPROVE OUTCOMES at the cost of error rates or profits

    Prikaži ovu nit
    Poništi
  4. 31. sij

    They assume that data sources are group-specific and infinite, and market segments have no distributional similarity and must train models separetely on each group. Each firm has a profit maximisation problem, potentially subject to regulatory constraints.

    Prikaži ovu nit
    Poništi
  5. 31. sij

    The setup is that AI market consists of firms developing models and concumers. Consumers are divided into non-overlaping groups i.e. market segments, they choose firm based on the model performance. Firms profit=market share - costs on data collection.

    Prikaži ovu nit
    Poništi
  6. 31. sij

    Catching up with , who presented a fascinating work at AIforGood on data markets, economics and pac-learning, its implications on AI ethics and regulation! Check out his paper and read below for more details!

    Prikaži ovu nit
    Poništi
  7. proslijedio/la je Tweet
    30. sij

    I swear it's not just men working , but who doesn't love a boyband? They're fairly feminist too ⚡️ We also have and

    Poništi
  8. 30. sij

    A boy band⚡🎸 from , working at the edge of Earth Science AI! And generally, a nice bunch of people!

    Poništi
  9. 24. sij

    This “correlates” with the results in presented at workshop “that tech. experts are more optimistic than healthcare domain experts regarding current potential for automatability of healthcare work”.

    Prikaži ovu nit
    Poništi
  10. 21. sij

    However, as rightly points out - these decision trees are also opportunities for amazing research and clinically relevant applications! – mjesto: Royal College Of Pathologists

    Prikaži ovu nit
    Poništi
  11. 21. sij

    Excellent remark by . Decision making in clinical practice is an intricate tree (lung cancer tree in photo)! Frequently, AI publications oversimplify these work flows - smth that pointed out during ml retrospectives at

    Prikaži ovu nit
    Poništi
  12. proslijedio/la je Tweet

    In our data-driven world, the claim that we don’t have a good way to study something quantitatively may sound shocking. The reality even worse — in many cases we don’t even have the vocabulary to ask meaningful quantitative *questions* about complex socio-technical systems.

    Prikaži ovu nit
    Poništi
  13. proslijedio/la je Tweet

    I'd argue the reverse too. You can outperform in pretty much any scientific field by bringing with you an artistic mindset.

    Poništi
  14. proslijedio/la je Tweet

    Not even catastrophes like these seem to bring any political action. How is this possible? Because we still fail to make the connection between the climate crisis and increased extreme weather events and nature disasters like the That's what has to change. Now.

    Poništi
  15. proslijedio/la je Tweet

    WOMEN 👏🏼 WHO 👏🏼 DON'T 👏🏼 WANT 👏🏼 KIDS 👏🏼 • aren't selfish • are not less of a woman • aren't immature • won’t necessarily change her mind • won’t necessary be a "lonely old cat lady" which is a disgusting patriarchal stereotype • don’t owe you any explanation whatsoever

    Prikaži ovu nit
    Poništi
  16. proslijedio/la je Tweet
    14. pro 2019.

    So spot on! 'AI Ethics for Systemic Issues' by of ! We all really need to be considering and talking about the more complex issues around in a broader sense.

    Slide showing the topics of Fairness, Privacy, and security of the tip of the AI Ethics iceberg that we need to consider.
    Poništi
  17. 14. pro 2019.

    Come hear speak on structural approach to AI ethics and the systemic impacts of AI at 11.55(East building 11) at and check out her cool poster!

    Poništi
  18. 12. pro 2019.
    Poništi
  19. 11. pro 2019.

    Is that a Rolling Stones concert? No, that's a keynote at

    Poništi
  20. proslijedio/la je Tweet
    10. pro 2019.

    More takeaways from this amazing work on algorithmic injustices by 1) Centre society's most vulnerable rather than the technology. 2) Automated decision-making systems create and sustain a certain social order. 3) Prioritise understanding rather than predictions.

    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:

    ·