2

During the self-play training process, AlphaZero does not greedily play only the moves it thinks are "best" (which would normally be the move with the highest visit count leading out of the root node of the MCTS search tree). Instead, for the purpose of generating a more diverse set of experience, it samples moves proportionally to the visit counts. This ...


2

As far as I understood, most RL applications have much more states than there are actions to choose from. Yes, this is quite common, but in no way required by the underlying theory of Markov Decision Processes (MDPs). The most extreme version of the opposite thing - with one state (or effectively no state, as state is not relevant) - are k-armed bandit ...


2

In RL, the phrase "acting greedily" is usually short for "acting greedily with respect to the value function". Greedy local optimisation turns up in other contexts, and it is common to specify what metric is being maximised or minimised. The value function is most often the discounted sum of expected future reward, and also the metric used when defining a ...


2

In general, a greedy "action" is an action that would lead to an immediate "benefit". For example, the Dijkstra's algorithm can be considered a greedy algorithm because at every step it selects the node with the smallest "estimate" to the initial (or starting) node. In reinforcement learning, a greedy action often refers to an action that would lead to the ...


2

According to the book Artificial Intelligence: A Modern Approach (3rd edition), by Stuart Russel and Peter Norvig, specifically, section 3.5.1 Greedy best-first search (p. 92) Greedy best-first search tries to expand the node that is closest to the goal, on the grounds that this is likely to lead to a solution quickly. Thus, it evaluates nodes by using ...


1

You are calculating the average reward for each action (i.e. bandit arm) incorrectly. You cannot calculate this simply with a list comprehension, and you need to keep a second list storing the number of times each action was taken. The correct calculation is to divide the total reward obtained from each action by the number of times that action was taken. ...


1

There are two factors that will change the ability of a deep neural network to fit a given dataset: either you need more data, or a deeper and wider network. Since the pattern is only 2-d, it can likely be approximated by some sort of simple periodic function. A DNN can approximate periodic functions pretty well, so the issue is probably that you don't have ...


1

Shannon's mouse was obsessed with the cheese, which was an obsession with winning too, just not winning by prevailing over another mouse. There was no other hungry mouse in the maze. There is a ring of truth to the idea that, as AI develops, individual AI systems will become more like minds. Artificial minds would need a healthy growing environment that ...


1

What you said isn't totally wrong, but the A* algorithm becomes optimal and complete if the heuristic function h is admissible, which means that this function never overestimates the cost of reaching the goal. In that case, the A* algorithm is way better than the greedy search algorithm.


Only top voted, non community-wiki answers of a minimum length are eligible