We’ve discovered that including adaptive noise to the parameters of reinforcement studying algorithms steadily boosts efficiency. This exploration manner is unassuming to put into effect and really hardly decreases efficiency, so it’s value making an attempt on any downside.
