AlMeta:YTDNM reinforcement learning
Your the Dog Now Ma(n|chine)
- if (you got a reward)
- if reward is > 0
- Treat =true
- add top of @History to @Trick
- else (if reward < 0)
- Treat = False
- clear @Trick
- for X in @History
- update X in NN
- if (X was Conditioned)
- Reward = Reward minus( 1 minus likelihood of choosing X)
- if(X was not Conditioned)
- Reward = Reward times 0.5
- clear @History
- if (you are not sure what to do)
- if (Treat)
- if (second of @Trick not Conditioned)
- add @Trick to @Repertoire
- if (second of @Trick not Conditioned)
- add new choice to @History as not Conditioned
- if (Treat)
- else (if you are sure what to do)
- add new choice to @History as Conditioned
- if reward is > 0
- else (if you didn't get a reward)
- if (your in not vary sure of what do)
- if (you last action was Conditioned)
- reward = -1
- while(top of @History was Conditioned)
- pop top of @History as X
- Update X in NN
- reward = reward times likelihood of X
- Clear @History
- Clear @Trick
- add new choice to @History as not Conditioned
- if (you last action was Conditioned)
- else (if you are sure of what to do)
- if you last action was Conditioned)
- add top of @History to @Trick
- add new choice to @History as Conditioned.
- if you last action was Conditioned)
- if (your in not vary sure of what do)