• Neuroscience & Pharmacology Discussion Welcome Guest
    Posting Rules Bluelight Rules
    Recent Journal Articles Chemistry Mega-Thread
    FREE Chemistry Databases! Self-Education Guide

  • N&PD Moderators: Skorpio

Artificial Neural Network for Drugs Anyone

L

lariyn

Bluelighter
Joined
Sep 26, 2005
Messages
45
Ok this may seem weird. But it does relate to drugs. I am training a network to tell people what drugs to do. It learns when ever you decide to do a drug, you just tell it all the information and tell it what you thought about it. It also learns when you ask what drug to do and give all the information and if it is right you reward it and if its wrong you punish it. Would anyone actually be interested in playing with such a application. anyways I also wanted to know what do people use to judge what drugs to do, like time of day, mood, energy, sleep, and im not sure what other things to include.

BTW its a Cyclic Custom Artificial Neural Network using Unique Confidence Bounds and Activation Function per Neuron and Propagation Rule per Network using Fuzzy Logic Implemented using Genetic Algorithm. The Confidence Bounds and Activation Function can be changed per Neuron, and the Propagation Rule and Training Algorithm can also be changed in at any time for the entire network. It is cyclic so the output feeds back into the input, giving this amazing network a concept of time, so it takes into account all previous drugs injested as long as you tell it what you did it trains it self.

anyway if anyone cares go ahead and respond. i dont expect really anyone to care except me. later and have fun all
 
Last edited:
Sounds intriguing. I'd like to do some more reading on neural nets and information theory and get back to you some day, but I'm not sure when that'll be. I do care; it's just a hell of an investment of time in studying to prove it ...
 
yeah only reason im doing it is because it sounds interesting. As far as working, who knows its ganna really be some wierd ass training. But it will probably just tell you to do crack all day.
 
so the goal is to come to a conclusion about what decisions people actually make?

seems likely that at least 99% will come from past experience.
 
that sounds insanely sweet, I didn't quite understand everything you mentioned, since i know very little about the programming stuff. but this is really relevant to me, every day i have to decide what to take and what not to take, and similar stuff like that. It would be really cool to have some way to input all the variables , and have something that would keep track of everything and analyze it......maybe i would stop making the same mistakes over and over again, i have a terrible memory:)
 
i already said this, but i read it again, And it sounds even awesomer. Usually ( im completely serious) i hold 2 different substances , one in each hand, and i make up a modified version of Iny miny miny moe, or flip a coin. Specifically, for me im taking the stuff to alleviate Mental disorder problems, and I don't always know which Pill will help me the most at that specific time. I try and think back to my previous experiences, but ALWAYS there is something I forget, until later when its no longer as important.

this sounds like a really cool idea
 
Fuck why not? What do we have to do? obviously I have way to much free time on my hands or I wouldn't be reading this in the first place.
 
ok this is why ethics committee's exist.

Let me pose a hypothetical question if this was a pharmaceutical company would you let THEM get away with this experiment?
 
Phener said:
ok this is why ethics committee's exist.

Let me pose a hypothetical question if this was a pharmaceutical company would you let THEM get away with this experiment?
Click to expand...
Would they actually provide drugs for me to choose from? If so I'd have to say yes. Repeatedly.
 
when I was young, a boy a grade ahead of me had the official Biker meth hook up.
his mum's side of the family was sending it to cny upstate from Cali.
i was to be impressed by the biker sub-culture motto of "whatevr those fucking losers can manage to import into the community I was living in at 17yo."

I was brainwashed into being an addict

if your travels ever cross into the 315...................................
SRS, Pm me first.
edu. does not impress me
Im for real, you are advocating drugs and a venture in the human mind for dispersal.

anyways......i like to jerk off on stimulants

315......you hear me????????
 
Implimentation

atm i just finished the custom recurrent artificial neural network and am finishing designing the input notations used, and designing the output layer as well. I have started experimenting with the training process, as well as the neuron count and connections to find the best network configuration. When designing the network great care needs to be taken as one to large will take forever to train and might not work as good as one with the least possible neurons and layers cause less neurons and a better configuration will train faster and have better results. I am also experimenting with activation functions, confidence intervals, propagation rules( wanna see how well competitive propagation works ) once i feel it has reached a suitable point i will post a link to download and try this out. did not get really any responses at first so was about to abandon the project for the moment, but if there are at least some people interested then ill continue with development.
 
neural network

btw a neural network is what your brain is. its neurons connected to other neurons sending signals like your brain. it learns by two different training algorithms atm, im using reinforement( punish and reward ) for predicting the drug, and backpropagation for telling it what drug to do corresponding to certain input conditions. also experimenting with genetic algorithms for training. have a net that every individual neuron can configure with different confidence intervals and activation function, plus the network can switch training algorithms and propagation rules. made a net that can configure to many different configurations so i hope experimenting goes smoothly as i do not have to reprogram a new net with different configurations, just switch out the configuration in the same net without any reprogramming.
 
The net will not converge, your input is underconstrained. Feel free to continue to waste your time however.
 
your prob right ima have to think carefully redesign the input layer, like the data used for predicting. i see what you mean thats mainly why i was looking for some ideas. but yeah to get it working even remotely im going to have to do alot of experimenting. finishing a vb app, with all the ann code and related code from a dll coded in c++ and asm, to configure the network so once finished experimenting will hasten. the app for running the net and training it is all in c++ and asm using a home brewed gui, i think looks sweet so no messing around with the console, i dont know about you but that would be lame to the max. ganna try maybe also working with a recurrent configuration plus also im trying certain custom net configurations, so im trying nets thats not really a feedforward propagation network, but im also messing around with conventional feedforward propagation networks too, trying everything atm, and mess with the neuron count extensively as i have found too many it fucks training up and doesnt give good answers also two few and it cannot hold the pattern being trained so pretty much nonsense output. but yeah i totally agree, if i understand you correctly, that i need to decide on a better set of data for input. Im ganna start on that asap actually start simple, ganna suck though totally not ganna give good actually useful answers, and extend the input data and see what works and what doesnt. i dont plan on it being something to live by, as drugs really are not the way to go, but basicly some advice if you have no idea and still im really only hoping for the best cause i still think its really interesting app. if anyone has any advice i would really welcome it, a novel perspective is always useful.
 
Last edited:
It learns when you tell it when you do a drug and what the initial conditions were at that time to let it learn your pattern of use. If it learned some pattern then you can give it some data about some initial conditions and it will attempt to give an idea of what drug you might like to do. Aslo you can punish it or reward it depending on if it is correct.
 
You must log in or register to reply here.
Top