So the sigmoid function creates a limit of 1 as x approaches infinity meaning big high heapum activation value gets closer and closer to 1 ?

At a guess it only ever reaches 1 because of precision
Limitations ?

  • 0
    God kill and fuck with every silky hipped chick you can throw at me for the next x years you bastards !
  • 0
    20 < x < 120
  • 0
    Fuck you lol
  • 0
    Looks like I have to learn past this to have a new question god I hate this
  • 1
    Sums up what is left heh
  • 0
    When is the logistic sigmoid function ever 0 ?
  • 1
    Reaching 1 or 0 exactly: never, it limits to them at infinities.

    Reaching 1 or 0 "close enough": pretty fast. Plug in some numbers and see for yourself. Getting the sigmoid to exactly 1 or 0 is pretty useless.

    The point of the sigmoid is that it's a differentiable ("smooth and unbroken") replacement for the step function (aka the "hard threshold" implemented by x < 0.5 ? 0 : 1). Being differentiable is useful for optimization methods like the backpropagation used in neural networks because it makes the problem amenable to gradient based optimization algorithms (like stochastic gradient descent and its friends).
  • 0
    @RememberMe that’s the thing though all the docs suggest these functions are outputting both

    Maybe the tutorials are misleading with their examples

    I still err on the side of thinking it’s a rounding or precision thing I get the confidence index idea

    The huge activation value hitting the output would create a value ever approaching 1 smallll values would make e^-x approach 1 which would result in a sum that is greater than one which would make sigmoid into a smaller decimal value

    Still the value only has a fuzzy meaning for logic

    How the hell donsome machines output coordinates ??

    Or do they not does the implemhatioj feed it partial image fragments ?
  • 0
    @RememberMe that’s kind of the part not being clearly defined
    But I see this one example of dnn in opencv that takes a whole video frame and returns a copy with multiple sometimes overlapping boxes classified

    Albeit sometimes correctly and sometimes not

    Is it sort of side feeding partial resized images into it and reading the inference output values and expanding a bounding box until a high classification output is not returned ?
  • 0
    @RememberMe and in which case where does it decide to start ? or how to scale image pieces ?

    like a picture of an umbrella at exactly 250,250 might be different than one that is 50,50 in the same image,showing one closer and another farther away

    but if the expected image size is 100,100 for the inference model... how in the hell is dnn using modelnet even working as well as it is ?? especially given its classifying multiple objects (or sometimes misclassifying them) AND detecting them !

    is it two different parts like i'd think ? image segementation, resize and then inference ? or do they actually find a way of feeding the whole image or during training just put random shit scaled at different sizes to a horizon line ?
  • 0
    @RememberMe since you know supposedly its classifying pixels. as belonging to a group.
  • 0
    @RememberMe seriously btw some help finding one of the following:

    1. people who are simply not chomos at all
    2. people who are not in positions of success because they can stand being around garbage scum all the time and really don't care whats going on in the world at all

    would be nice.
  • 0
    @YouAllSuck You okay buddy? Every rant I see you post or comment in goes to strange new places.
  • 0
    @lungdart well lets look at it this way.
    everytime someone asks me that question, in that exact format, its the same as a standardized form of speech chomos from after my father's generation would ask it because they view people as either individuals who were tormented as children and grew to angry extremely rageful adults that despise them, or grew into them. and they ask that and there is some hidden suggestion in there as if they are secretly hoping the answer will be an unpleasant one. the answer is as it was previously in response to this exact response is quite simple: if i'm not doing so well, is IT doing all that great ? because if we are not doing well, than IT must be doing the same shit over and over not allowing US to live and be happy and have to put up with ITS dumb chomo ass.

    but the short answer is, I'm exxasperated over how fucking retarded half the country is right now and was just thinking yet again about how sad it is that you can't convince the young...
  • -1
    @lungdart not to be idiots like your dumb fucking ass likely is, thereby creating a world which may be happier for those like myself, while those like IT just slowly die off and do the world a favor and take their fucked up pygmy retard version of english with them, instead of poisoning every film, every movie, every bit of music, porn, whores, stripclubs, nude reviews, theater, technology, acamdemia, etc etc etc with their retarded self serving delusional and simplistic interpretation of literally fucking everything !
  • 0
    @YouAllSuck Talking about the U.S.? You guys seem to be working on some tough issues.

    Seems like years of propaganda is finally boiling over. If it makes you feel any better, it's coming to everywhere else, just a little lagging.
  • 0
    @lungdart for as much trouble as you fucking retards some of which are likely logging on from accounts in this very facility, are causing me, you fucking retards are stuck because you people CHOSE to make the world harder to live in instead of embracing progress and keeping your dick out of things that are beneath adulthood. :) that about answer it pastor ?
  • 1
    or whoever this is based on zim telling me a long time ago so i'd recognize people like to play the trick of signing up for accounts in my local environment just like online strippers for example are sometimes broadcasting on closed network circuits nearby indicative of a real person working with lots of time accounted for with repeat looped performances in a pay to talk channel which noone will pay for ;P
  • 0
    Heh looks like someone harassed with a deleted account and their comments went bye bye
Add Comment