Ryan J. O'Neil

@ryanjoneil

Real-time models and optimization. Founding . Integer scientist. PhD. Prefers . Formerly Grubhub, Zoomer, MITRE, Washington Post.

Traveling
Vrijeme pridruživanja: srpanj 2013.

Tweetovi

Blokirali ste korisnika/cu @ryanjoneil

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

  1. Prikvačeni tweet
    13. ruj 2018.

    Hey look! 's and my post, "Decisions are First Class Citizens", is up on the blog!

    Poništi
  2. 27. sij
    Poništi
  3. 27. sij

    Yeah, that was ridiculous too. Somehow the TSP stuff seems worse to me.

    Poništi
  4. 27. sij

    Is the same attention to detail paid to stories on climate change? What about coronavirus? I don't know much about these things, and I'd like to have faith that what I read has been researched. This is as far as anyone has to go to find the state of the art in TSP solving.

    Poništi
  5. 27. sij

    What bugs me in these TSP articles: They're not just a little bit wrong. They're light years off from decades-old reality. Lots of MS students study TSP solving. These aren't the dark arts. If the news is so wrong about this, how can I believe articles about anything technical?

    Poništi
  6. 27. sij

    And on and on. I forgot that and got on the bandwagon for that one too. Surely someone at one of these establishments would like to correct the record?

    Poništi
  7. 27. sij
    Poništi
  8. 27. sij

    Any journos out there want to write a story correcting this rubbish? published an article just as bad a couple years back. did the same thing. It's gotten pretty shameful.

    Poništi
  9. proslijedio/la je Tweet
    26. sij

    2. Maybe better tooling for to reach more people without heavy experience on discrete algorithms. The solvers we use are always comparing each other in terms of constraints implemented and performance. How about usage? Time to setup a project? Documentation?

    Prikaži ovu nit
    Poništi
  10. 24. sij

    Long live Integer Science! If I weren't too busy right now, I would totally apply.

    Poništi
  11. 18. sij

    It gets worse when you consider online shoppers who don't own cars, request delivery, and purchase large boxes. "No, thanks, I just ordered 40 lbs of cat litter."

    Prikaži ovu nit
    Poništi
  12. 18. sij

    From a consumer perspective, the state of recommendation systems often seems to be "you bought cat litter, would you like to buy more cat litter? Here are 15 other brands of cat litter."

    Prikaži ovu nit
    Poništi
  13. 13. sij

    To all the developers who struggle with naming things... You are not alone.

    Poništi
  14. 12. sij

    I thought it was when you switched to fitness proportionate search, but neglected to hit the gym.

    Poništi
  15. 12. sij

    Stumbling into the abyss, essentially. 😁

    Poništi
  16. proslijedio/la je Tweet
    12. sij

    I wonder if it is true that myopic is more used when talking about stochastic optimization. For instance, a myopic policy ignores future rewards and optimizes direct rewards. On the other hand, greedy is more used in classic combinatorial optimization (like Dijkstra) Not sure...

    Poništi
  17. proslijedio/la je Tweet
    12. sij
    Odgovor korisnicima

    that does contradict my def. If you check a textbook def, greedy needs the "local optimal" while myopic is just the opposite of "far sighted", well literally, or "not global", it is weaker than greedy. If both were the same, why would one proof properties for greedy algos?

    Poništi
  18. 12. sij

    Interesting alternative view re: greedy vs myopic.

    Poništi
  19. proslijedio/la je Tweet
    12. sij

    That being said, to asnwer the question, I'd say that I use both terms interchangeably

    Poništi
  20. 12. sij
    Poništi
  21. proslijedio/la je Tweet
    11. sij
    Odgovor korisniku/ci

    I would use “greedy” when referring to a choice that neglects existing data (eg nearest neighbor tsp heuristic) and “myopic” for choices made before new information is released over time (eg maximizes today’s profit before seeing tomorrow’s demand data). Time aspect is key

    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:

    ·