user avatar
user avatar
user avatar
DaVinci
  • Member for 6 years, 6 months
  • Last seen more than 3 years ago
4 votes
Accepted

How is my definition of "information" different than Shannon's entropy?

4 votes

Q-values in a DQN are spiralling out of control

3 votes

When implementing TD(λ) with eligibility traces, why do we update towards the same 𝛿?

3 votes
Accepted

Backpropagating a Dueling Architecture Network: Gradient Calculation

3 votes

Why do we have to use action-value function in model-free reinforcement learning instead of just state-value function?

2 votes
Accepted

Why don't we update when exploring (making non-greedy moves)?

2 votes
Accepted

Understanding oscillating behaviour when using Q-learning on cart-pole problem

2 votes

Why does training naive Bayes on a data set in which all the features are repeated increase the confidence of the naive Bayes probability estimates?

2 votes

Decision making in different intervals in MDPs

1 vote

Minibatching in Stochastic Gradient Descent and in Q-Learning

1 vote

Why is the discount term in the REINFORCE algorithm never used?

1 vote
Accepted

Parameter or Network Switching?

1 vote

How best to dynamically update online decision trees weighting?

0 votes
Accepted

Transition matrix for continuous state-action space reinforcement learning