2 ms·
Your depression theory has some resemblance to reinforcement learning. An agent learns to estimate the value of any state and to predict future states. When the
by visarga 9y ago
Your depression theory has some resemblance to reinforcement learning. An agent learns to estimate the value of any state and to predict future states. When the value function is corrupted, it might get stuck in a very bad local optima. The way to escape is to try more things (to do random actions that might take you out of the loop). In RL, there needs to be a balance between exploration and exploitation. Depression seems to be a problem of too little exploration.