People glad to be back! I innovated a branched technique in psychology, to form symbol tables. Its call decision trees. Basicly it gathers input and tells you what drugs to do when any not to do. Works off of artificial intelligence tecniques expored by me recently. How many people out there want there phone telling them what drugs to do all day? We should build a database on decisions and responses regarding ingesting drugs!
-
N&PD Moderators: Skorpio | thegreenhand
artificial neural networks for drugs anyone II
- Thread starter scrut.hut
- Start date
yoyoman
Bluelighter
- Joined
- Jun 11, 2006
- Messages
- 304
i've had the same thought!
I just don't know anything about coding neural networks.
But to feed it input, might take a lot of human work or maybe not... maybe run "good drugs" through by running every good drug you can think of. Am I right you can input this any way you want like an IUPAC name or SMILES, or structure image, then input a whole bunch of bad drugs or very neurotoxic drugs. I'm curious what the output would be. If you made a network for good stimulants and ran all stimulants through would it be able to out put a novel new good stimulant?
I just don't know anything about coding neural networks.
But to feed it input, might take a lot of human work or maybe not... maybe run "good drugs" through by running every good drug you can think of. Am I right you can input this any way you want like an IUPAC name or SMILES, or structure image, then input a whole bunch of bad drugs or very neurotoxic drugs. I'm curious what the output would be. If you made a network for good stimulants and ran all stimulants through would it be able to out put a novel new good stimulant?
You'd probably need a lot of processing power to get anything useful from this but it could be fun.
Is this parallelizable? If it is and you wrote an open source program that would run on anyone's cpu/gpu there'd probably be enough processing power to get something useful from it. As for the inputs, like the people before me said, train it on known good & bad drugs and then see what it comes up with....
Is this parallelizable? If it is and you wrote an open source program that would run on anyone's cpu/gpu there'd probably be enough processing power to get something useful from it. As for the inputs, like the people before me said, train it on known good & bad drugs and then see what it comes up with....