David Zoltowski

@DavidZoltowski

Statistical neuroscience, machine learning. PhD student

Princeton, NJ
Vrijeme pridruživanja: lipanj 2009.

Tweetovi

Blokirali ste korisnika/cu @DavidZoltowski

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

  1. Prikvačeni tweet
    16. sij

    Excited to share what and I have been working on recently! We unify and generalize statistical models of neural dynamics during decision-making using switching state-space models

    Prikaži ovu nit
    Poništi
  2. 16. sij

    Finally, code is available at for the models and at for vLEM

    Prikaži ovu nit
    Poništi
  3. 16. sij

    I'll be presenting this at Cosyne 2020, so if you're at the conference and interested you can come by my poster to hear more!

    Prikaži ovu nit
    Poništi
  4. 16. sij

    To fit the models, we developed a scalable inference method for (r)SLDS models called vLEM. vLEM is a generalization of Laplace EM (see , 2011) for single-state LDS models.

    Prikaži ovu nit
    Poništi
  5. 16. sij

    We demonstrated the approach on two sets of spiking recordings from monkey parietal cortex during decision-making, where we compared 1D vs. 2D accumulators and fit a ramping model with a variable lower bound.

    Prikaži ovu nit
    Poništi
  6. 16. sij

    This provides a framework for modeling neural dynamics with 1D accumulation to bound dynamics (e.g. the DDM), multi-dimensional accumulators, variable & collapsing boundaries, trial-history effects, and more!

    Prikaži ovu nit
    Poništi
  7. 7. pro 2019.

    Normalizing flows were recently used in comp neuro by Sean Bittner and colleagues for identifying how theoretical neuro model parameters relate to emergent properties of interest

    Prikaži ovu nit
    Poništi
  8. 7. pro 2019.

    Check out this great video introduction to normalizing flows by !

    Prikaži ovu nit
    Poništi
  9. 23. lis 2019.

    Check out this awesome work from and colleagues! They developed a model-based targeted dimensionality reduction method that when applied to PFC recordings identifies accumulation followed by rotational dynamics in a multi-dimensional subspace.

    Poništi
  10. 30. srp 2019.

    Belated but huge congrats to ! Anqi's done a lot of awesome work applying GP latent variable models to analyze neural recordings (and more). We'll miss you in Princeton!

    Poništi
  11. proslijedio/la je Tweet
    22. srp 2019.

    Excited to share a new preprint: "Discrete Object Generation with Reversible Inductive Construction", with , , Abigail Doyle, .

    Prikaži ovu nit
    Poništi
  12. 22. srp 2019.
    Poništi
  13. proslijedio/la je Tweet
    4. lip 2019.
    Poništi
  14. 3. lip 2019.

    This will surely be fun. will this be available online?

    Poništi
  15. 3. svi 2019.

    This awesome work is leading the push to develop increasingly powerful models of neural data that identify interesting and interpretable structure. Congrats Scott!

    Poništi
  16. proslijedio/la je Tweet

    3 years ago MIT grad student Katie Bouman led the creation of a new algorithm to produce the first-ever image of a black hole. Today, that image was released. More info: 2016 story:

    Prikaži ovu nit
    Poništi
  17. proslijedio/la je Tweet
    2. tra 2019.

    “If sexual harassment, misconduct, and retaliation are the firing squads that assassinate individual careers, then implicit bias is the lead in the water that poisons the entire town.”

    Prikaži ovu nit
    Poništi
  18. 15. ožu 2019.
    Poništi
  19. 8. ožu 2019.

    Spoiler alert: the inductive bias of oi-VAE helps it learn much less painful representations of walking than a VAE. Nice work from Sam Ainsworth, Emily Fox, and others!

    Poništi
  20. proslijedio/la je Tweet
    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:

    ·