Alec Tschantz

@a_tschantz

PhD student with the Sackler Center for Consciousness Science.

Vrijeme pridruživanja: veljača 2018.

Medijski sadržaj

  1. 22. ruj 2019.
    Odgovor korisnicima

    The conclusion of the Laplace approximation is that you can replace the environment variables in F with the mean of your recognition distribution. With the Dirac delta, this is as simple as (taken from the Bogacz tutorial):

  2. 24. lip 2019.
    Odgovor korisnicima i sljedećem broju korisnika:

    I agree that the "value" should refer to the probability of a policy, but the figure shows deontic value as the likelihood of an observation given a policy, no?

  3. 24. lip 2019.
    Odgovor korisnicima i sljedećem broju korisnika:

    The equation says "the log likelihood of an observation given a policy", whereas the words says "the likelihood of a policy given an observation"

  4. 24. sij 2019.
    Odgovor korisnicima i sljedećem broju korisnika:

    I think that the simplest derivation from free energy would be this (where 'epistemic' scores the information gain and 'reward' scores the agreement between expected & prior (i.e. favourable) beliefs):

  5. 7. sij 2019.
    Odgovor korisnicima i sljedećem broju korisnika:

    example use of the API:

  6. 7. sij 2019.

    Indeed you have - compare equation 3 from your paper (attached in different notation) to equation 3 of

  7. 11. ruj 2018.

    "After analysing the OpenAI Five network it becomes clear that most of the network is dedicated to perception (preprocessing the observations) and motor control (decoding the actions). All the strategy and tactics must lie in one place – 1024-unit LSTM."

  8. 9. kol 2018.
    Odgovor korisnicima
  9. 4. kol 2018.
    Odgovor korisniku/ci

    The framework espoused by Clark & others (predictive coding) relies on the same objective function as variational auto-encoders:

  10. 18. srp 2018.
  11. 13. srp 2018.
    Odgovor korisniku/ci
  12. 8. tra 2018.
  13. 8. ožu 2018.

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:

    ·