Hottest Questions This Month

4

votes

2

answers

275

views

Given two optimal policies, is an affine combination of them also optimal?

modified Nov 20 at 14:13 nbro♦ 22.2k

1

vote

1

answer

242

views

How is Google Translate able to convert texts of different lengths?

natural-language-processing recurrent-neural-networks machine-translation google-translate seq2seq

modified Nov 10 at 11:38 razvanc92 948

2

votes

2

answers

212

views

If uniform cost search is used for bidirectional search, is it guaranteed the solution is optimal?

search proofs uniform-cost-search bidirectional-search optimality

modified Nov 20 at 15:18 Community♦ 1

2

votes

1

answer

67

views

Why is the completeness of UCS guaranteed only if the cost of every step exceeds some small positive constant?

search uniform-cost-search completeness

modified Nov 9 at 23:26 Nathan S. 331

1

vote

2

answers

78

views

How to verify classification model trained on classification dataset on a detection dataset for classification purpose?

deep-learning convolutional-neural-networks object-detection image-processing

modified Nov 16 at 22:19 Douglas Daseeco 7,310

1

vote

1

answer

73

views

Why does Monte Carlo policy evaluation relies on action-value function rather than state-value function?

reinforcement-learning monte-carlo-methods value-functions model-based-methods model-free-methods

modified Nov 20 at 14:09 calveeen 1,046

3

votes

1

answer

70

views

Why does off-policy learning outperform on-policy learning?

reinforcement-learning comparison q-learning off-policy-methods on-policy-methods

modified yesterday nbro♦ 22.2k

3

votes

1

answer

68

views

What is the difference between Stochastic Hill Climbing and Simulated Annealing?

comparison simulated-annealing meta-heuristics local-search stochastic-hill-climbing

answered Nov 23 at 20:56 nbro♦ 22.2k

2

votes

1

answer

66

views

Are there any meaningful books entirely written by an artificial intelligence?

natural-language-processing applications natural-language-understanding natural-language-generation

modified Nov 2 at 12:19 Oliver Mason 3,860

1

vote

1

answer

80

views

Is the expected value we sample in TD-learning action-value Q or state-value V?

reinforcement-learning monte-carlo-methods temporal-difference-methods value-functions return

modified Nov 23 at 13:37 nbro♦ 22.2k

1

vote

1

answer

128

views

What is the space complexity of breadth-first search?

search breadth-first-search branching-factors space-complexity

modified Nov 9 at 19:09 nbro♦ 22.2k

1

vote

2

answers

76

views

Why not use the target network in DQN as the predictor after training

reinforcement-learning dqn deep-rl

modified Nov 1 at 12:00 Otniel Mercea 39

0

votes

1

answer

87

views

What is the optimal value function of the shifted version of the reward function?

reinforcement-learning proofs bellman-equations reward-shaping reward-functions

modified Nov 2 at 13:47 nbro♦ 22.2k

1

vote

1

answer

105

views

What is the space complexity of bidirectional search?

search branching-factors breadth-first-search space-complexity bidirectional-search

modified Nov 9 at 19:25 nbro♦ 22.2k

4

votes

1

answer

101

views

Why does a negative reward for every step really encourage the agent to reach the goal as quickly as possible?

reinforcement-learning proofs reward-shaping reward-functions

modified Nov 5 at 19:37 BlueMoon93 841

0

votes

1

answer

96

views

What is the space complexity of iterative deepening search?

search branching-factors homework space-complexity iterative-deepening

modified Nov 9 at 19:27 nbro♦ 22.2k

1

vote

2

answers

58

views

How to use sigmoid as transfer function when input is not (0,1) range in ANN?

neural-networks activation-function sigmoid

answered Nov 20 at 21:04 Alexander Soare 723

3

votes

1

answer

89

views

+50

How should we interpret this figure that relates the perceptron criterion and the hinge loss?

loss-functions support-vector-machine perceptron binary-classification hinge-loss

answered Nov 24 at 0:41 ombk 101

4

votes

1

answer

86

views

Can models get 100% accuracy on solved games?

machine-learning function-approximation

modified Nov 13 at 7:41 Neil Slater 15.9k

3

votes

1

answer

52

views

How are afterstate value functions mathematically defined?

reinforcement-learning value-functions bellman-equations afterstate-value-functions

modified Nov 24 at 16:48 nbro♦ 22.2k

2

votes

0

answers

85

views

+100

Generating fake faces containing specific features with GANs

machine-learning generative-adversarial-networks generative-model

modified Nov 24 at 19:28 MarekK 21

2

votes

1

answer

49

views

What exactly is the eigenspace of a graph (in spectral clustering)?

graphs clustering linear-algebra graph-theory spectral-analysis

modified Nov 8 at 19:01 pasaba por aqui 1,073

1

vote

1

answer

49

views

What kind of problems is DQN algorithm good and bad for?

neural-networks reinforcement-learning dqn applications

modified Nov 22 at 20:03 nbro♦ 22.2k

1

vote

1

answer

59

views

How do I create a custom gym environment based on an image?

reinforcement-learning open-ai gym a3c

answered Nov 23 at 15:24 Ren 21

3

votes

0

answers

74

views

+50

In GradCAM, why is activation strength considered an indicator of relevant regions?

neural-networks deep-neural-networks explainable-ai grad-cam

modified Nov 23 at 13:33 nbro♦ 22.2k

0

votes

1

answer

57

views

What is the meaning of these equations in Noise2Noise paper?

papers supervised-learning notation expectation mean-squared-error

modified Nov 23 at 10:56 nbro♦ 22.2k

1

vote

0

answers

73

views

How should I build this DQN agent?

reinforcement-learning deep-learning tensorflow keras dqn

asked Nov 25 at 19:51 GeezLouiz 11

1

vote

2

answers

44

views

Why is an embedding of dimension 400 enough to represent 70000 words?

recurrent-neural-networks hyperparameter-optimization word-embedding hyper-parameters sentiment-analysis

modified Nov 7 at 13:40 nbro♦ 22.2k

2

votes

1

answer

56

views

When learning off-policy with multi-step returns, why do we use the current behaviour policy in importance sampling?

reinforcement-learning off-policy-methods value-functions importance-sampling return

modified Nov 18 at 6:29 Shreyans Jain 11

1

vote

2

answers

71

views

Why don't neural networks project the data into higher dimensions first, then reduce the size of each layer thereafter?

neural-networks overfitting multilayer-perceptron generalization

modified Nov 24 at 15:41 nbro♦ 22.2k

1

vote

1

answer

43

views

Should we also shuffle the test dataset when training with SGD?

machine-learning training datasets stochastic-gradient-descent testing

modified Nov 3 at 13:57 nbro♦ 22.2k

2

votes

1

answer

42

views

Is it possible to retrieve the optimal policy from the state value function?

reinforcement-learning dynamic-programming

answered Nov 7 at 13:09 Neil Slater 15.9k

1

vote

1

answer

42

views

Why aren't exploration techniques, such as UCB or Thompson sampling, used in full RL problems?

reinforcement-learning multi-armed-bandit upper-confidence-bound thompson-sampling exploration-strategies

modified Nov 16 at 23:18 nbro♦ 22.2k

-1

votes

1

answer

67

views

How to extract the main text from a formated text file?

neural-networks

modified Nov 24 at 17:10 AlexSC 99

0

votes

1

answer

51

views

Keras DQN Model with Multiple Inputs and Multiple Outputs [closed]

convolutional-neural-networks python tensorflow dqn multilayer-perceptron

answered Nov 15 at 13:57 yang2x 26

1

vote

1

answer

38

views

Is which sense was AlphaGo “just given a rule book”?

reinforcement-learning alphazero alphago-zero alphago muzero

modified Nov 7 at 0:44 nbro♦ 22.2k

1

vote

0

answers

62

views

Does the order in which the features are concatenated to create the state (or observation) matter?

reinforcement-learning deep-rl feature-selection state-spaces observation-spaces

modified Nov 20 at 16:58 mark mark 147

1

vote

1

answer

48

views

AI to play a solo card game

game-ai ai-basics monte-carlo-tree-search monte-carlo-methods minimax

modified Nov 1 at 14:00 Neil Slater 15.9k

3

votes

0

answers

60

views

In Soft Actor Critic, why is the action sampled from current policy instead of replay buffer on value function update?

reinforcement-learning policy-gradients value-functions off-policy-methods soft-actor-critic

modified Nov 23 at 1:43 nbro♦ 22.2k

3

votes

0

answers

60

views

Has the logistic map ever been used as an activation function?

neural-networks activation-function

modified Nov 24 at 10:23 nbro♦ 22.2k

1

vote

1

answer

46

views

Why does the accuracy drop while the loss decrease, as the number of epochs increases?

neural-networks overfitting loss accuracy epochs

modified Nov 5 at 11:28 nbro♦ 22.2k

3

votes

1

answer

45

views

How does MuZero learn to play well for both sides of a two-player game?

reinforcement-learning self-play muzero

modified Nov 1 at 13:43 Ziofil 150

1

vote

1

answer

45

views

Can the (sparse) categorical cross-entropy be greater than one?

convolutional-neural-networks training object-recognition categorical-crossentropy alexnet

modified Nov 19 at 12:07 nbro♦ 22.2k

2

votes

1

answer

57

views

Is this referring to the true underlying distribution, or the distribution of our sample?

machine-learning classification probability-distribution selection-bias

answered 2 days ago Dennis Soemers♦ 7,980

1

vote

1

answer

44

views

If the reward function of an environment depends on some initial conditions, should I create a separate environment for each condition?

reinforcement-learning environment reward-design reward-functions

modified Nov 25 at 9:10 nbro♦ 22.2k

1

vote

1

answer

33

views

What is the impact of changing the crossover and mutation rates?

genetic-algorithms hyperparameter-optimization crossover mutation genetic-operators

modified yesterday nbro♦ 22.2k

2

votes

1

answer

32

views

Is it possible to pre-train a CNN in a self-supervised way so that it can later be used to solve an instance segmentation task?

transfer-learning r-cnn self-supervised-learning instance-segmentation pretext-tasks

modified Nov 20 at 16:56 nbro♦ 22.2k

1

vote

2

answers

32

views

Why is my Keras prediction always close to 100% for one image class?

deep-learning computer-vision keras bayesian-deep-learning uncertainty-quantification

modified yesterday nbro♦ 22.2k

Top Questions

Looking for more? Browse the complete list of questions, or popular tags. Help us answer unanswered questions.