Here, you can simulate a possible future performance of different machine learning algorithm on given number of
arms (drugs). Patients are assigned sequentially to drugs. In other words, at each step, one patient is assigned to a drug by the algorithm. We also
assume the effect of the drug (reward) is seen immediately. The goal here is to make sure more patients are assigned to the optimal drug.
To setup the experiment, first, you need to answer the following question. By default, some values are previously assigned and can be changed. The hyper parameters of algorithms can also be changed to see their effects on performance.
Then, select number of drugs (arms) and change their effect distribution (reward distribution). The drug that has higher reward is better. Finally, please press the submit button to see the corresponding simulation graphs. Currently we support bernouli distributions for drug effects (reward distribution). In the future, we will add more distributions.